This action might not be possible to undo. Are you sure you want to continue?

989

**Modeling and Simulation of Contacting Flexible Bodies in Multibody Systems
**

by

Iakov Nakhimovski

Submitted to the School of Engineering at Linköping University in partial fulfilment of the requirements for the degree of Licentiate of Engineering Department of Computer and Information Science Linköpings universitet SE-581 83 Linköping, Sweden Linköping 2002

**Modeling and Simulation of Contacting Flexible Bodies in Multibody Systems
**

by Iakov Nakhimovski Dec 2002 ISBN 91-7373-559-0 Linköpings Studies in Science and Technology Thesis No. 989 ISSN 0280-7971 LiU-Tek-Lic-2002:62 ABSTRACT This thesis summarizes the equations, algorithms and design decisions necessary for dynamic simulation of flexible bodies with moving contacts. The assumed general shape function approach is also presented. The approach is expected to be computationally less expensive than FEM approaches and easier to use than other reduction techniques. Additionally, the described technique enables studies of the residual stress release during grinding of flexible bodies. The overall software system design for a flexible multi-body simulation system BEAST is presented and the specifics of the flexible modeling is specially addressed. An industrial application example is also described in the thesis. The application presents some results from a case where the developed system is used for simulation of flexible ring grinding with material removal. This work has been supported by SKF Sweden AB, KK-stiftelsens företagsforskarskola i Linköping and the ECSEL (Excellence Centre for Computer Science and Systems Engineering in Linköping) Graduate School.

Department of Computer and Information Science Linköpings universitet SE-581 83 Linköping, Sweden

.

associate professor in Numerical Analysis at Lund University. I am grateful to all the administrative staff at IDA. in particular Lillemor Wallgren. ECSEL (Excellence Center for Computer Science and Systems Engineering in Link¨ ping) Graduate o School and KK-stiftelsen f¨ retagsforskarskola in Link¨ ping. The work was ﬁnancially supported by the SKF Sweden AB. A special thanks goes to Bodil Mattsson-Kihlstr¨ m for handling all the adminiso trative work caused by my frequent travels to Gothenburg. Theirs u comments helped to signiﬁcantly improve the text. Special thanks to Lars-Erik Stacke for proof reading the thesis.Acknowledgments Many people contributed to the work presented in this thesis and essentially made the thesis possible at all. H˚ kan B˚ stedt. I would like to thank my parents in Russia. Leonid Gershuni. November 2002 o . encouraged me to tackle the problems during the whole period of studies. First. Many thanks also to all my colleagues at PELAB who have created a nice working atmosphere at the University. Iakov Nakhimovski Link¨ ping. and by o Claus F¨ hrer. Furthermore. and Alexei Jolkin a a as well as Daniel Lindgren from Xdin for many fruitful discussions and lots of useful feedback. the head of PELAB (Programming Environments Laboratory) at Link¨ ping University. I really appreciate the reviews of the thesis done by Peter Fritzson. Lars-Erik Stacke. and Bodil Carlsson. Link¨ ping University deserves a special thanks for o helping me in understanding of some of the mathematical problems discussed in the thesis. and gave many important comments on the text of the thesis. o o Finally. Britt-Inger Karlsson. and my relatives and friends in different countries around the world for constant moral support and belief in my ability to do the work and write this thesis. I also want to thank all the other members of the BEAST team at SKF: Mikael Holgersson. Many thanks go to my fellow industrial PhD students at SKF Alexander Siemers and Aleksandar Bogdanovski for sharing the exciting experience of studying and working on BEAST at the same time. I would like to thank my supervisor Dag Fritzson at SKF Sweden AB who has suggested many of the ideas realized in this work. Fredrik Berntsson from MAI.

.

. . . .5 External Point Moment . . . . . . . . . . .Contents Notation 1 Introduction 1. . . . . . . 1. . . . . . . . . . . . . . . . . .1 Flexible Body Motion and Shape Functions . . . . . . . . . . . . . . . . . 2. . . . . . . . . . 2. . . . . . . . . . . 2. . . . .4 Special Shape Functions for Solid Bodies in Cylindrical Coordinates 1 3 . . . .4 Quadratic Velocity Vector . . . . . . . . . . . . . . . . . . . . . . . 2. . . 2. . . . . .4 Related Work . . . . . . . . . .3 Model Requirements . . . . . . . . . . . . . . . . . . . . . . . . .5. . . . . . . . . . . . . . . . . . . . . . . . . . 2. . . General Shape Functions 3. .5. . . . . . . . . . . . . . . . . . .5 Thesis Overview . . . . .2 Generalized Newton-Euler Equation . . . . . . . . 4 8 8 9 12 14 15 17 17 19 20 22 23 23 24 27 27 28 29 30 31 34 34 36 40 40 41 42 44 . . . . .5. . . . . 1. . . . . . . . . . . . . . . . . . . . . . 3.4 External Point Force . . 2. . . . .1 Motivation . . . . . . . 2. . . . . . . .3 Separation of Elastic and Rigid Body Motion Modes . . . 3. . . . . . . . . . . . . . 1. . . . . . . .8 An Interpolation Method for Forces with Discontinuities .2 Choice of Shape Functions . . . . . . . . . .7 Using Intermediate Coordinate System . . . . . . . . . . . . .5. . . . . . . . . . . . . 2. . . . 2. 2. . . . . . . 2 Key Equations 2. .5. . .1 Generalized Elastic Forces . . . . .3 Generalized External Forces . 2. . . . . . . . . . 3. . . . . . . . . . . . 2. . . . .6 Rotation of a Material Particle due to Deformation . .5. . . .6 External Volume Load . . . . . . . . .2 The BEAST Toolbox . . .3 Mass Matrix . . . . . . . . . . . . . . . . . . . . .2 Generalized Viscosity Forces . . 2. . . . . .5. . . . . . . . . . . . . . . . . . . . . . . . . . . .8 Calculating Jacobian . . . . . . . . . . . . . . . .1 Simulation of Mechanical Systems 1. . . . . . . . . . . . . . 2. . . . . . . . . . . . . . . .5.5 Generalized Forces . . . . . .7 Generalized Forces from the Internal Stress Release during Grinding . . . . . . . .

3 Object-Oriented View of Shape Functions . . . . . . . 7. . . . . . . . . . . .5. . . . . . . . 7. . . . . . . .6 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 Flexible Body Class Design . . . . . . . . . . . . .3 Grinding Machine Model . . . . . . . .5 Grinding simulation results . . . . . . . . . . .1 System Overview . . . . . . . . . . . . . 7 Simulation System Design 7. . . . . . . . . . . . . . . . . . . . . . . . . 8. . . . . . . . . . . . . 45 46 48 51 51 53 54 56 57 57 58 60 60 61 63 63 64 66 66 66 68 68 69 70 70 72 73 73 73 74 75 76 81 82 84 85 4 Modeling Ideal Connections with Control Points 4. . . . . . . . 4. . . 7. . . . . . . . . . . . . . . . . . 8 Dynamic Simulation of Grinding 8.3. . Eigenmode and Quasi-static Single Body Analysis 5. . . . . . . . . . . . . .3. . . . . . . . . .2 Implementation of the Control Point Architecture 7. . . . . . . . . . . . . . . . . . . . . . . 5. . . . . . . . . . . . . . Reducing the Number of Flexible States . . .2 Rigid Body Model Classes . . . . . . .2 Jacobian Evaluation . . .2 Need for Grinding Simulation . . . . . . . . . . . . . . . . 8. . . . . . . . . . . . . . . 7. . . . . . . . . . . .4 Modeling Tribological Contacts with Material Removal 8. . . .1 Control Points and Flexible Bodies .2 Boundary and Loading Conditions . . . .1 Static Loading Cases .1 Long Beams . . . . . . . . . . . . . 7. . . . . . 5 Static.3 Quasi-static Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7 Imposing Constraints for Rigid Body Motion . . . . . . . . . . 9 Conclusions 10 Future Work A Shapes Generated by General Functions 2 . . . . . . . . . . . . 4. . . . . . . . . . . . . . .5 3. . 8. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 An Example of Control Points Usage . 6 Veriﬁcation 6. . . . . . .1 Overview . . . . . . . . . . 8. . . . . . . . . . . . . . . . . 7.4 A Simple Model Example . . . . 7. . . . . .3. .1 RHS Evaluation . . . . 6. . . . . 5. . . . . 7. . . . . . .1 Limitations of the Rigid Body Model Design . . . . . . . . . .5 Solution Procedure . . . . . . .2 A Thin Ring . Volume Integration . . . .3 Flexible System Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7. . . . .3. . . . . .5. . . . . . . . . . . . . . . . . . . . .6 Summary . . . . . . .3.2 Eigenmode Analysis . . .

B Example Acceleration Calculations for a Flexible Ring Bibliography 88 103 3 .

the basic quantities and some formulas are given in both notations. The notation used here is both the direct matrix notation of vectors and tensors. T T a second-order tensor or matrix. T . For further reading we refer to [9. Therefore. in some cases an array. That is a matrix: The important properties of a skew symmetric matrix are: normally a second-order tensor or matrix. and the vector-dyadic notation.Notation The notation used throughout the thesis is described below. a skew symmetric matrix constructed from the components of 3-element vector . 8] or other well-known standard publications. 4 transpose of 1 % ¥¦ ¡¦ & ¥ ¦ & % '¦ $ !§ ¨ ' 0)( ¡ ¦ ¦ & % "# 3 §2 4§ ¨ §2 § ¨ § ¨ & !§ ¨ §¨ ¨ §¨ 7 © 987¦ © © ¦ © ¥¤¤¢ £¡ ¨ @ §¨ 6 ¦ 5 § ¨ §¨ ¨ ¦ transformation matrix from coordinate system 1 to system 2. 11. 7. The section can be used as a reference when reading the equations presented in the thesis. a scalar. a unit vector. The orthonormal right-handed rectangular Cartesian coordinate system is used. and . Variables & basic deﬁnitions a vector or ﬁrst-order tensor on which vector transformation rules can be applied.

moment. damping matrix.e. ff 1/c ext 1.e. . ff ff E . ext . refering to object 1 a force couple. The subscipts specify the degrees of freedom (DOF) corresponding to the particular sub-block.2 sub-blocks of complete mass matrix . a control point deﬁned in the coordinate system unit base vector in a rectangular Cartesian axis system Euler parameters elastic (Young’s) modulus. matrices of elastic and viscous material constants. moment. stiffness matrix. length or width a moment relative to the origin of system c refering to object 1 an external force couple. i. i. time derivative of a vector with respect to coordinate system c. ’ ’ specify the three translational DOFs. c If no coordinate system is speciﬁed then the global inertial coordinate system is assumed. refering to object 1 and caused by interaction with object 2 complete symmetric mass matrix of a body as used in NewtonEuler equations. ) dot product of two vectors dot product of two matrices . a external force refering to object 1 an identity matrix moment of inertia for object 2 relative to coordinate system 1 some of the inertia shape integrals deﬁned by the equations 2-9. 2-10 and 2-13 respectively.(or cross product of two vectors a constant that gives the relative stiffness between object 1 and object 2. The nondiagonal 5 `dQ e c ¥ ¤E Q T § I 3 9§2 W ¤ © DS7R07© P W Q © 97R07© P Q SR7 P 7Q @ @ vW y@ vwW u ¥ § t xt ¥£ ¤¤¡ t he @ fe @ g ' ¡e@¥e © c a` ¥ ¤E bQ9VQ U U V3 §I 3 §I §I 3 G GHEFC C B § ¨ D@ A § ¨ G c Y ¥ § r p qi i ¡ ¥ X Y 2 !¨ 2 !§ ¨ §2 !§ ¨ §2 § Y ¥ § § §¨ s ¦ (c) a 3-element vector expressed in component form in coordinate system ’c’. ’ ’ the three rotational DOFs and ’ ’ elastic DOFs.

blocks of the matrix use mixed subscripts. components of the deﬂection vector f(b) in the direction of coordinate axis . 6 gk@ j@ i f h h d h g k ~ @ z g {@ z}|@ f ` @y f §e d W t u h k j h c i h ¥¤¤¡ § s £ rq© § l o p§ l §l g @ ng @ g m §lf §lf §lf Wg © w@ © vW x t xt vH§ t xw W §ef §e §l § § t §e . a “point”. volume. a dynamic inertia shape vector deﬁned by the Equation 2-15. ’2’. rotational and elastic DOFs respectively.e.2. mass and mass density of object . ’3’ in different equations as appropriate in the particular context. quadratic velocity tensor.g. Axes are named ’ ’. generalized force due to internal residual stress release. that is the row associated with the -th Cartesian coordinate axis. The subscript is omitted if the discussion clearly indicates a single object. that is a 3 matrix representing the deformation ﬁeld of a ﬂexible body. the location of the origin of coordinate system 2 relative to the origin of coordinate system 1 a shape matrix. nine inertia shape integrals deﬁned by Equation 2-12. velocity. a/b (b) f (b) f an element of the point vector associated with the axis . e. If b is a coordinate system. stands for the sub-block deﬁning inertia coupling between the translational and rotational DOFs. a ”point” vector expressed in the coordinate system b. The subscripts and specify corresponding coordinate axes. one of the inertia shape integrals deﬁned by Equation 2-11. ’ ’ or ’1’. i. total generalized force tensor as appears on the right hand side of the Newton-Euler equation deﬁned in 2. f parts of a generalized force vector corresponding to translational.. the vector between a and b. f the tensor of the generalized elastic coordinates of a ﬂexible body. is used to denote the -th row of the matrix. ’ ’. then it refers to the origin of system b. Where is one of or for a Cartesian coordinate system and one of for a cylindrical coordinate system. external generalized force tensor.

viscose stress tensor. d B ¡ ¤£ § ¥ @ ! @ a potential energy. ¤¤¡ § ¥£ w@ § § § © ¤£ § ¥ a ¥£ ¤¤¡ § 7 . angular acceleration of ’a’ relative system 1 differenciated in system 2. strain energy and work of external forces. rotation angle. elastic stress tensor. Poisson’s ratio.elastic strain tensor. This tensor cannot be expressed in a coordinate system. -th vibration mode frequency [radians/second]. If no system 2 is speciﬁed then system 1 is assumed. inﬁnitesimal rotation angles. the angular speed of coordinate system 2 relative to coordinate system 1 . a ﬁrst order tensor containing three angles representing the rotation of coordinate system 2 relative to system 1. the angular speed of “a” relative to coordinate system 1. Lame’s constants.

Some measurements in a real dynamic system are very hard or even impossible to perform. Tuning the system parameters is much easier for a simulated system than in the real machine. Many machines may be dangerous to use without simulation in advance. Multi-body dynamics simulations typically deals with systems of interconnected rigid bodies.Chapter 1 Introduction 1.1 Simulation of Mechanical Systems Simulation technology becomes more and more important to industry. On the other hand. Sometimes manufacturing of a small slightly modiﬁed part at a factory or workshop requires several weeks. Mechanical systems are obvious candidates for simulation for many reasons. e. Simulation nowadays is a must. two main types of simulation of mechanical systems have been dominating: Building a prototype mechanical system just to test a new design is very expensive and time consuming. However in a simulation all the variables are accessible. and more detailed simulations on a computer.g. Some particular effects that in real systems are obscured by some other phenomena can be isolated and carefully studied in a simulation. some of them are listed here: Historically. cheaper. Simulation systems can be seen as virtual test rigs that allow for replacement of long and costly experiments on a real test rig with faster. in other situations such small effects can be completely neglected to allow more careful study of the main relations. In many cases the connections are limited to idealized 8 . an airplane or car design.

. The general development of simulation software goes in the direction of simulation of complete systems with more and more detailed and accurate analysis. and discusses the object-oriented design of the computer implementation.. Multi-body systems are mostly used for time domain dynamic simulations. This means that multi-body systems now include deformable bodies and more detailed analysis in the models. This thesis contributes to this process on the multi-body systems side.joints (e. They are also used for detailed simulations of very small and speciﬁc parts of a system. . At the same time ﬁnite element tools are beginning to be used for the dynamic simulations. Here. numerical analysis. It discusses the ﬂoating-frame of reference formulation which is gradually becoming the standard way to simulate ﬂexible bodies in the context of a multi-body simulation. An example of such a system is the BEAST (BEAring Simulation Tool) described in the next section. provides some important numerical algorithms. prismatic.2 The BEAST Toolbox BEAST (BEAring Simulation Tool) is a fully three-dimensional simulation program developed at SKF and originally used to perform simulations of bearing dynamics 9 Finite element tools are mostly used for static and quasi-static (with constant angular velocities) analysis of systems and components as well as for modal analysis of structures. Finite element analysis generally requires more computational effort than multi-body analysis. and computer science are necessary. spring. a single contact. The toolbox is used industrially at SKF . The thesis discusses the necessary mathematical model for deformable bodies simulation. Therefore all aspects of the system development have to be considered.. e. An industrial simulation system is always the result of interdisciplinary research where contributions from different ﬁelds are integrated into a single generally useful system. To further narrow the ﬁeld of the research we limit our development to multi-body systems specializing in detailed contact analysis. mechanics provide the mathematical models of physical phenomena in the form of equations. numerical analysis comes up with the methods and algorithms for the solution of the equations. damper. etc) and traditional force elements (e.g. The aim of this thesis is to describe an approach for building a ﬂexible multibody simulation system as an extension to the existing rigid body one (BEAST).g. revolute. Multiple dynamic contacts are often important parts of the model and a relatively simple contact model is most often employed for performance reasons. 1. external load).g. and computer science gives the design guidelines for the efﬁcient implementation of the model on a computer. In the case of ﬂexible multi-body simulation system contributions from mechanics.the world’s leading rolling bearing manufacturer.

BEAST can now be seen as a general multi-body simulation tool specialized in detailed contact analysis. This has enabled studies of internal motions and forces in a bearing under any given loading condition. Some of the features of the tool include animating the simulated sequence from different viewpoints. visualization of force vectors and surface associated data (such as pressure distribution). transmissions. ViewBeast is specially designed for 2D plot presentation and analysis. The visual representation of the simulation results in Beauty in an easily understandable way contributes to a quicker interpretation of the result and popularity of the complete toolbox among the users. In addition 10 . BEAST has been seen as a virtual test rig where the user has full insight into the dynamic behavior of the bearing components. Beauty Input Animations Output ViewBeast 2D plots Output Output Input Input Output Input RunBeast Output Input BEAST Out2In Figure 1-1: Programs in the BEAST toolbox. The tools are: Beauty is an advanced 3D graphical tool designed for setting up the model and simulation parameters in the input ﬁles and visualizing the output ﬁles. Later versions of BEAST include models of grinding machines.on any bearing type. The bearing could be loaded in any way the user required. experimental engines. etc. The toolbox development in terms of design generalization has widened the application area of the program. The BEAST toolbox includes a set of tools that are interacting with each other by means of different kinds of ﬁles. Figure 1-1 shows the programs in the toolbox and their interactions with the ﬁle storage. Figure 1-2 shows a snapshot of the Beauty working on an input ﬁle for a grinding machine model.

It reads in a model speciﬁcation from input ﬁles. This means that is the results of one 11 . RunBeast is a remote simulation interface system. curve plotting. Most of the developments of this thesis were realized within the framework of this tool. Its architecture is described in [22]. i. modify some parameters.. BEAST is the main simulation program. For example. and continue the simulation from the moment where the previous run was terminated. different operations on the simulation results are possible. performs the simulation and generates a set of output ﬁles containing the results.Figure 1-2: A snapshot of the Beauty displaying a grinding machine model. to the basic functionality. the user can apply Fourier transforms to a variable or specify his/her own function on several variables. The tool provides a user-friendly interface for submitting a simulation on a remote computation server which is often a parallel computer.e. In this way the user is given a chance to interrupt the simulation. Out2In is a small utility that allows generation of new input ﬁles from the simulation output ﬁles.

The most important features of this solver are: The original solver has been extended to handle special kinds of Jacobians (block diagonal with borders and sparse) and to perform a simultaneous RHS and Jacobian calculation [13].3 Model Requirements The main aim of this thesis is to provide the elastic body model suitable for dynamic simulations involving multiple moving contacts. and computationally expensive evaluation of the derivatives. Adaptive time step changes based on a local error estimate. From the mathematical point of view the BEAST system solves the simultaneous system of Newton-Euler equations of motion for every body in the mechanical system.8 and [13] for more details. Its successful industrial usage validates the correctness of the decisions taken.e. They are: the computational performance or efﬁciency.simulation can be used as the initial conditions for another. All the ideas for the modeling. and the design decisions were tried for applicability and usability in this system. The tool uses a modiﬁed version of the CVODE [4] differential equation solver for the numerical solution of the resulting system of equations. the matrix of partial derivatives of the state derivatives with respect to the state variables). For the work described in this thesis. 12 A variable order multi-step method that requires high continuity of the derivatives. the accuracy. Efﬁcient use of the solver requires a fast procedure for the system Jacobian calculation (i. the algorithms. Local deformation in a contact is covered by the contact model discussed in other reports on the BEAST (BEAring Simulation Tool) and GRIT(GRInding simulation Tool) systems [6. 23. The main issue is an efﬁcient model of the overall structural deformation. . the potential to incorporate the analysis needed for grinding. The second order differential equations system is rewritten as a ﬁrst order system.. Typical characteristics of such ODEs are: mathematical stiffness. This solver is one of the most well known free implementations of the implicit backward differentiation formulas (BDF) integration scheme. the BEAST system has been the assumed implementation platform. The Newton-Euler equations are formulated as second order ordinary differential equations (ODEs) on explicit form. 12]. Hence we can say that the system has become a test implementation of the presented approaches. The requirements on the model can be subdivided into three equally important groups. See Section 2. 1. very high numerical precision of the solution required by the application.

That is why an efﬁcient ﬂexible body model for the case of moving contact should not force the numerical solver to take signiﬁcantly smaller time steps compared to the same model with rigid components. In Figure 1-3 a rotating elastic ring pressed between two plates is shown. Let us provide a very simple example to illustrate the need for accuracy. due to the initial residual stresses in the surface layer combined with uneven material removal. the higher frequencies in the system are expected to be associated with the contact forces calculations and not with the ﬂexible body eigenfrequencies. Figure 1-3: A rotating ﬂexible ring pressed between two plates. In the third group of requirements one can mention distortion. measured as the distance between the plates. The situation can get worse for a more complex model where small. The system of ODE (ordinary differential equations) that is generated in the dynamic contact analysis problems is in most cases mathematically stiff. Longer time steps means less computations and shorter simulation times. Hence. The computational complexity of a ﬂexible system model depends mostly on the number of state variables used for the ﬂexible body. it is important that we do not introduce vibrations due to modeling errors or numerical errors. That is. an efﬁcient model should use as few states to describe ﬂexibility as possible. should become constant after some transient process. Important consideration for such solvers is the length of the time step in the integration. When simulating this in the time domain. induced vibrations can occasionally resonate with some resonance frequency of the system leading to completely unrealistic results. From the physical point of view the radial stiffness of the ring is independent of its orientation. The commonly used type of numerical integrator for such systems is multi-step variable time step solver. Hence assuming the constant force acting on the ring the elastic deﬂection. 13 .

The required procedures are described in a wide variety of articles and books. However. It is one of the most complicated and discussed problems in ﬂexible multi-body simulation (see [20]). Reduction schemes often dramatically decrease the number of state variables used in dynamic simulation. 30] from a detailed ﬁnite element model. For our simple example described above. The result of a reduction procedure is a set of assumed displacement shape functions that is used in the dynamic simulation. Also. weeks) computation times. i.4 Related Work A 3-D FEM model can be used to model ﬂexible bodies. However. References [1. 5] provide extensive discussions of the methods. A commonly used approach to efﬁciently simulate structural elasticity is to employ some kind of reduction technique [5. in the case of a moving contact the boundary conditions are changing dynamically. These resulting mode shapes can be calculated only for some speciﬁc boundary conditions.3. due to the discrete interface nodes on the contact surfaces... 14 .e. variation in radial stiffness leads to induced non-physical vibration with the frequency dependent on the number of elements on the circumference of the ring and rotational speed. Interaction between any two points of the contacting surfaces is possible. This can be done directly in the reduction scheme [30]. The effects of the discretisation errors in dynamic simulations are discussed in [16]. The solution to this is to employ continuous shape functions for the interface motion.e. The main challenge of a reduction with modal synthesis is to select a small set of superimposed mode shapes that give acceptable calculation precision for the simulation. the radial stiffness computed at the element nodes will slightly differ from the stiffness between the nodes. This is due to the fact that the deﬂection between ﬁnite element nodes is normally computed using a low order interpolation scheme. The mode shapes are calculated using data from static mode shapes and/or frequency response modes. or may be done as a separate transformation step after a standard reduction with all interface nodes. i. 29. given attachment or interface nodes. Without doubt the effect can be minimized and made negligible by using a large number of ﬁnite elements but that would lead to an even longer computational time. ﬁnite element methods for transient analysis often suffer from long (days. Large number of interface nodes leads to the large number of state variables and makes the dynamic simulation slower. control the surface interface nodes by the shape functions.. non-linear effects due to initial stresses and material removal.1. These reduced models cannot handle the requirements from the third group mentioned above in Section 1. and eigenmode shapes. The reduction is done prior to the simulation using a FEM-tool. i.e. The reduced model still has the same vibration problem as the FEM model. Hence many nodes on the surfaces have to be marked as interface nodes.

The complete derivation of these equation can be found in most books on ﬂexible dynamics (see [21]). These inertia shape integrals. The second chapter introduces the mathematical model of an elastic body using the ﬂoating frame of reference formulation and assumed shape function.1. There are several kinds of analysis that are possible only for ﬂexible body models. Appendix B shows how the equations presented in this chapter can be used for modeling of a ﬂexible ring. The third chapter presents the set of general shape functions that are proposed to be used as assumed mode shapes for elastic bodies of used in simulations with detailed contact analysis. The thesis is structured into seven chapters and two appendixes. There is also a discussion on a more specialized kind of analysis: ring distortion due to the residual stress release during grinding. The set of the inertia shape integrals required for the equations is deﬁned. Results of these kinds of analysis can be used to quickly assert the basic properties of the ﬂexible body model before running computationally heavy dynamic simulation. This thesis summarizes the equation of motion for deformable bodies that undergo large translational and rotational displacements using the ﬂoating frame of reference formulation. The ﬁrst chapter provides an introduction and an overview of the thesis. appear in the non-linear terms that represent the inertia coupling between the reference motion and the elastic deformation of the body. body deformation under static load and under quasi-static conditions. The necessary requirements and boundary conditions as well as volume integration procedure applicable to the presented 15 . On the other hand. The thesis is therefore system implementation oriented and does not generally cover the mathematical derivation of the presented equations. that depend on the assumed displacement ﬁeld. All the important equations that deﬁne the model are listed and explained in this chapter. This thesis describes three general types of such analysis: free-body eigenmodes analysis. the dynamic formulation of the system of equations of motion for linear structural systems requires the deﬁnition of the system mass and stiffness matrices as well as the vector of generalized forces. and the generalized forces acting on the body. The thesis also provides some examples that were used to verify the correctness of the mathematical model presented. the inertia tensor.5 Thesis Overview The dynamic equations of motion for the rigid bodies in a multi-body system can be deﬁned in terms of the mass of the body. Some numerical procedures required to perform a dynamic simulation involving ﬂexible bodies including numerical volume integration and Jacobian matrix evaluation are also described. The main intention of this thesis is to provide a complete and possibly compact description of the equation and procedures necessary to perform a dynamic simulation of a ﬂexible multi-body system and some simpler kinds of single body analysis.

Chapter four covers some design decisions that were made to provide an efﬁcient user interface to for specifying boundary and external loading conditions in simulations involving ﬂexible bodies. 16 . Chapter seven focuses on the system architecture and design for a multi-body simulation system that includes ﬂexible components. We specially address the problem of transition from a multi-body system that allows only rigid bodies to the system with both kind of components. Appendix A provides a complimentary example for this chapter.shape functions are discussed in this chapter. The ﬁfth chapter describes some extra types of analysis that are possible for elastic body models. Chapter eight describes an application example for the simulation system discussed in the thesis. Chapter six gives some simple example test cases that were used to verify the accuracy of the ﬂexible body modeling using ﬂoating frame of reference formulation and general shape functions. The last two chapters summarize the results of the performed work and discuss the possible directions for the continuation of the research and development in this area. The importance of the structural ﬂexibility and its inﬂuence on the simulation results for the particular application are discussed.

Appendix B shows how the equations presented here can be used for modeling of a ﬂexible ring. 2. 17 Representation in the different coordinate systems. Jacobian calculations.Chapter 2 Key Equations This chapter introduces the mathematical model of an elastic body using the ﬂoating frame of reference formulation and assumed shape function. Explicit step-by-step calculation procedure as presented in Appendix B. On the other hand.the global and body ones . The chapter is strongly based on the results of [21] and therefore no derivation for the most of the presented equations is given. Generalized external moment. . Since the equations involve only two coordinate systems .1 Flexible Body Motion and Shape Functions A rigid body in space has six degrees of freedom that describe the location and orientation of the body with respect to the ﬁxed frame of reference. Generalized force from the residual stress release. Generalized viscosity forces and damping modeling. The relations that are developed in this thesis and cannot be found in the reference above are: In the following sections the subscript that normally indicates the body number will be omitted with the understanding the all vectors and matrices are associated with some particular single body. Generalized external body load. An interpolation method for forces with discontinuities.one corresponding letter (’g’ or ’b’) will be used to specify the coordinate system where the vector is expressed or differentiated.

. The coefﬁcients . and f is the vector of elastic coordinates that contains the time dependent coefﬁcients . plates. the global position of an arbitrary point of the body can be deﬁned as where (g) b/g deﬁne the origin of the body reference.structural components such as beams. where where f(b) gives the displacement of an arbitrary point that has coordinates 0(b) in the undeformed state. The above equations can be written in the following matrix form: (b) f f where is the three-rows -columns shape matrix whose elements are the functions . By using the outlined approximations.dependent. and shells have an inﬁnite number of degrees of freedom that describe the displacement of each point on the component. . However. where . (b) (b) is equal to 0(b) f 0 in the body coordinate system Note that it is generally possible to specify the shape matrix in different coordinate systems and transform it using rotation matrix and normal rules for coordinate transformation. use of the body reference frame enables the most natural and efﬁcient choice of the shape functions. Hence the coordinate system needs to be speciﬁed for all the tensors in the following sections. For this reason the shape matrix is always assumed to be calculated in the body coordinate system. Since it is computationally impossible to deal with inﬁnite spaces. and are assumed to depend only on time. where . and .and time. classical approximation methods are employed wherein the displacement of each point due to elastic deformation is expressed in terms of a ﬁnite number of coordinates: (b) f (b) f (b) f . and . 18 ª ª (b) P/b § 3 t ª (g) P/g (g) b/g (b) P/b (2-3) f and gives the displacement of the point © §e §e § ¦ g § e f F¢ WF¢ W ¤¤¤¤ g § e f W W ¤¤¤¥ £¤ g § e f W W (b) 0 (b) 0 (b) 0 W zQ ¬£« § e 3 p W P ¦ W § 3 t W zQ W¢W FfzQ ¥ W 9hW P ¥ W W¦ ¥ §s §e W P W¦ § § W¡ ¤g W¢ y¤g vW ¤g §e §e §e §ef §ef §ef ¨¢ W §e W 9 @ W t §e © §e ¬w«® £ §s (2-1) (2-2) . The total number of elastic coordinates (which is of course equal to the number of columns of the shape matrix ) should be determined heuristically depending on the simulation accuracy required and the importance of the elasticity effects for the complete model. The vector of displacement (or deformation vector) f(b) is space. orthogonal rotation matrix.

is the angular acceleration vector of the reference of the deformable body deﬁned in the body coordinate system.2 Generalized Newton-Euler Equation The generalized Newton-Euler equation for the unconstrained motion of the deformable body that undergoes large reference displacement is given by: Alternatively using body coordinate system for all vectors: (b) b/g (b) f ff where b/g is the acceleration of the offset of the body reference coordinates relative to the global coordinate system. .orthogonal rotation matrix. second derivative of the vector of time-dependent elastic generalized coordinates of the body f . deﬁned for the rotation about on angle as (2-8) 19 g e y ¡ f & g @ 1 ¡ gg he ¥ e g he ¡ e () g e y ` g he e ¥ f & g g ¥ ' &ª¡ fe ¡ e f y ¡ fg' e f y & ' fe ¥ e f y ge' hfe ª g ' g g fe ¡ e f y g he ¡' e g ¥ e f y & ` g hezfe ' & ¡ ¡ e ¥ e f y ¡ fe f y & 6 ·¸ @ ·¸ @ 5·¸ · ¸ g 8¶¥ e ¶ ' ª& fee ¥ ee f yy $ ¬£ w« g¡ ¥ f ¡ ¡ e f y & ` "# § f 1 m§ l $ ² 4§ l ² )) () µ³µ³¬¬ 4§ l "### ) () e Q d | E °¢ j ¯ $ 1 §± § $ 1 yx t }yx t § ± s "## ) () x t vT § t "## "## Where g g 1 ²m g o § l f $ ª 1 A § 3 U & ²m § g 3 & § l f $ 1 ´l§²m l §lf o §lf µ³¬ ² § ) () µ³¬ ² g o § l f "## ) () ²¬µ³ g "## )) () w« ´§ l ³ w« ³w« § l f ³ Q d | E e°¢ j ¯ $ 1 §± Hx t § $ 1 yx t ¬ « ¬£« § ± s "## ) () x t £w® vT § t p "## (g) b/g (b) f ff R R f f ff f ff ff 1 m§ l $ ² ´§ l ² )) () w¬µ³³ « 4§ l "### R f R f ) () (2-4) # $ B §± s ¬£ w« ±§ § R (2-5) f (2-6) . which for the case of Euler parameters is deﬁned as: g (b) ½¹ ½ º Ä ¿¾»¡Ã¹ h·Â ÀÁ ghe @ q¼»¼»w ·· ¸ fe ¡q S¹§ ' º @ ¡ ½ w ¸ ¡ e @ ¡½ º ¹ ¸ ¥ e ~ where vector (2-7) are the four Euler rotation parameters.2.

where (b) 0 is the undeformed position of an arbitrary point on the deformable body. The mass matrix components now can be deﬁned: 20 ¿ - . The Newton-Euler equation given in this section can be further simpliﬁed by using the shape functions that satisfy mean-axis condition. 2. deﬁned by Equation 2-2. is the volume of the body. The components of the mass matrix are deﬁned in the next section. @y {z@ ` c @ h @ É v t W Y ²Ì Í Ç Æ W x t Ë Å v µ³¬ {z@ ` c @ h @ É v t WT t dÅ W x t @y Ç v É t ÊÇ ÅÁ x t @y {@ ` c @ h @ É v Y ²Ì Ë W Y ²Ì ËyÊÈÅ W u Ç v µ³¬ µ³¬ É e ÈÆ § Ç Å ¥§t (b) 0 (2-9) (2-10) (2-11) (2-12) (2-13) t s s É Î s Ç Ï { h W t §e .3 for further development and an efﬁcient solution approach. m . Note that in the special case of rigid body analysis the shape integrals are given by Equations 2-9 and 2-10 only. the deformable body inertia can be deﬁned in terms of a set of constant inertia integrals that depend on the assumed displacement ﬁeld.4.3 Mass Matrix Even though the mass matrix is highly nonlinear.is the identity matrix for the spatial case. where .The other parameters of the equation are the mass matrix. is the -th row in the body shape matrix . For the case of conservation of mass this matrix is the same for both cases of rigid and deformable bodies. See Section 3.mass of the body. the generalized forces and quadratic velocity vector. is the mass density of the body. and the quadratic velocity vector that includes the effects of Coriolis and centrifugal forces is deﬁned in Section 2.5. generalized forces are discussed in Section 2.

g g 3 1 g ¥'¥ ¡ xx tt && ¥¡¥' xx tt ff $ ª 1 g ¥'¥ ¡ xx tt && ¥¡¥' xx tt ff 3 g g () ¡ ' x t & ' ¡ x t f "# () ¡ ' x t & ' ¡ x t f 3 yx t g 3 WÓv x 3 T ª § t § § 3 g Ôvv x t v f 3 T § x t h ×c {yz@ ` c @ h Ö Õ @ 3 g W x t ª ÓWv x t ª ÓWv u & W } u v v ª § 3 g vÓv x v y f ª Óvv u v f W Y Ò u W Y } § t f Ò }yx t v § t - is given by § 3 x t ª ¥ t 6w' vY § 1 vY¥ % Ð& () ¡ vY Ð ' vY vY¥ Ð ¡ vY Ð $ ' vY Ð & % "# Ð % where Ð @ ¡ vY Ð @ vY¥ Ð 5 H§ t v - (2-17) Using the procedure similar to the one shown above we can come up with the following compact expressions for the diagonal and non-diagonal members of : }yx t Let us now show how this matrix can be calculated efﬁciently by using precalculated inertia shape integrals. First we will analyze the components of the diagonal elements: § 3 W x t 3 T § ª § 3 W qy ª W u W W xt W É ª § 3 6RW t 3 µ³W Y ¬ ²Ì Ë 5 Ç Î y ª É g 6É Î § 3 zW t 3 WT 5t Ç d3 T § ¡ µ³W Y ¬ ²Ì Ë f Ç Î É ª ª g 6 § 3 W t 3 µ³W Y ¬ ²Ì Ë 3 y § 3 W t 3 WT t 3 T § ¡ µ³W Y ¬ ²Ì Ë rf5 Ç Î É g ª 6 ¡ § 3 W t µ³W Y ¬ ²Ì Ë rf5 Ç Î É g 6 ¡ W Yµ³¬ £ ¬ ²Ñ Ë rf5 Ç Î É 1 )) 0))( $ ## "## is a g Ñ ª g Ñ Q d | E °¢ j ¯ e ¡ ¡ Yµ³¬ £ ¬ ² Ë f ¡ ¥ Yµ³¬ £ ¬ ² Ë f g ' Ñ ª g Ñ ¡ Y¬ ²Ñ Ë ' ²Ñ Ë ¡ Y³¬ µ£ ¬ ² Ë f ¡ ¥ Y¬µ³£ ¬ ² Ë f g µ³£ ¬ Yµ³¬ £ ¬ & ' Ñ ª g Ñ ¥ Y¬µ³£ ¬ ²Ñ Ë ' Yµ³¬ £ ¬ ²Ñ Ë & ¥ Yµ³¬ £ ¬ ²Ñ Ë ¡ Yµ³¬ £ ¬ ²Ñ Ë & ¡ Yµ³¬ £ ¬ ² Ë f ¡ ¡ Yµ³¬ £ ¬ ² Ë f ¿ symmetric matrix with the elements given by ÇÅ - Note that for the case of rigid body analysis the elastic coordinates are zero and so the matrix is constant. has three rows deﬁned as follows where 21 (2-19) (2-18) (2-16) (2-15) (2-14) T § $ TT §§ "# yx t .

4 Quadratic Velocity Vector The quadratic velocity vector can be deﬁned as In the three-dimensional analysis the components of the vector (b) (b) (b) where (b) is the angular velocity vector deﬁned in the body coordinate system.- is independent on the generalized coordinates of the body and. and is a skew symmetric matrix given by (b) (b) 3 (b) 2 (b) 3 (b) 2 (b) 1 The quadratic velocity vector includes the effect of Coriolis and centrifugal forces as nonlinear functions of the system generalized coordinates and velocities. constant: (2-20) 2. in terms of the shape integrals we need to evaluate the In order to express symmetric matrix: (b) 3 The part in the explicit form is given below: The term (b) 3 g g ª g § ª g § § § 1 ¡ § f § ¡ § § fwf & g g f ¡ § f § ¡ § § wf & g g § f § ª g § § § ¡ ¡ (b) 1 (b) 1 (b) 2 (b) 3 (b) 2 (b) 1 (b) 1 (b) 3 (b) 2 (b) 2 (b) 3 (b) 1 (b) 1 (b) 3 (b) 3 (b) 2 (b) 2 (2-24) f f (2-25) f can be further simpliﬁed to f (b) 2 f (b) 1 f (2-26) 22 () É ª ¦ }Ú6 A§ e 3Á 3 y ¤¤¤¤ g A § 3 x t f § & § 3 xA t & g 6 A§ x t y f (b) (b) f (b) f (b) 1 ¤¤¤¥ £¤ ª g ª g g ¥ A § 3 ' ¥ x t & ¥ ' x t f § A § 3 ¡ ' x t & ' ¡ x t f § A § 3 ¡ x t & ¡¥ x t f § A § 87© Ð 07© ¥ S' 7 ¥ r'© Ò ª g ¥ S7 ¥ r© g ¥ w7 ¥ r© A § 70© Ð 70© ' Û ' Û 3 y & § 3 87© x t T07© x t f 07¡© f ' Û ' Û & Ò p§ l o () 1 ' n' x t ª n¡ x t ª n¥ x t ¡ ¥ % § & § 6T % § § & T g o l f mT g o l f T g o l fr5 § § § % § & $ § z "# § g g Ù § e 3 ¡ Øf5 T t F Ç Î & o l f m¤g § opl f § 3 x t f § & ª g g § Hx t ¡ Øf5 3 w & w« ² op§ l f v ¬£« ³ o §l g opl f § § § wf & "# f g ¡ f g o¢l f § go lf § z $ (2-21) are deﬁned as (2-22) g ¡ f (2-23) . therefore.

Since the rigid body motion corresponds to the case of constant strains and since we deﬁned the deformation with respect to the body references.5. there is no loss of generality in writing the strain displacement relations in the following form: (b) f where is a differential operator deﬁned according to: or. introducing the material properties.5 Generalized Forces 2. .2. is the symmetric matrix of elastic coefﬁcients. Substituting Equation 2-31 into Equation 2-27 we can get: T f ff ff É g Å t Þ i T t Þf Ç Ï where 23 6 ' ¡ ¥ i T ' ¡ â @ ' ¥ â @ ¡¥ â @ n' â @ n¡ â @ n¥ 8â5 f {y@ ` @ g k z@ @~ }|f iC à ©7 @ d @ ¨w C Ý Ü ½ k C ª ~ CC | wC w ` | CC ª á k CC ½ w ¨w |½ w C C ª g w & ¨~ w C C f `| á kw CC g á¨w ª ½ ~ C w C f C `| ¨|á w C § Ü 3 É T § Ü ×§ Ç Å & § i W 7 Y ©¨w @ g 7 Y x¼w© Y xw ¥ Û ª © Y w ª 7 Y ¨w f `y W W 7 © ' 3 § & Þ § e Ï § 6 ' ¡ ¥ § T ' ¡ ß @ ' ¥ ß @ ¡¥ ß @ n' ß @ n¡ ß @ n¥ ß 5 § ! Ü Ý Ü n¡ ß ¡ ¥ n¥ ß '¡ß '¥ß ¡¥ ß ' n' ß § 07© ß § § Þ (2-27) (2-28) (2-29) (see [24]): (2-30) (2-31) (2-32) (2-33) . the constitutive equations relating the stress and strains can be written as where . and elastic forces.1 Generalized Elastic Forces The virtual work due to elastic forces can be written as is the virtual work of the where and are the stress and strain tensors. in case of cylindrical coordinate system with coordinates For a linear homogeneous isotropic material.

Boundary conditions discussed in Section 3.Poisson’s ration with the following equations: 1 % y ) % )) % )) )) () %% % % % Vy % % % Vy % ª % y % % % i 1 § 3 % & $ R ff f 0)( % % % ¤¤æ ä¤¤ å p A § qi "# % % % y ª § 1 m§ l $ ² ´§ l ² )) () w¬³µ³« ´§ l "### g ª f y ª g y ª f % % $ % ## # ## y ª "### i ¤æ ¤ ã ¤å ä¤ (2-34) can be expressed in terms (2-35) .ff is the symmetric positive semideﬁnite stiffness matrix associated with the elastic states of the body. For instance. for a linear homogeneous isotropic material. 2. accompanied by small strains holds.Young’s mod- (2-36) .2 Generalized Viscosity Forces We start the description by mentioning a big similarities between the elastic and viscosity forces and the related laws. The most important observation about these two measures is that the components of the tensors do not change under rigid body translation and rotation and so they provide a natural material description for the case of large rigid body motion but small elastic deﬂection and strain analysis. the constitutive equations relating the rate of strain change and viscosity stress can be written as (2-37) 24 g ª ` gy ` f ã & f g ` ª y ã f ã The Lame’s constants are related to the engineering constants ulus and . The resulting vector is therefore given by: f To complete the description we show how the matrix of Lame’s constants: It is important to mention that the stresses and strains we were describing above are the second Piola-Kirchhoff stress and Green-Lagrange strain tensors.3 ensure that the stiffness matrix becomes a positive deﬁnite one. The observation is of practical importance since Hook’s law is applicable only to small strains and because in the particular problems we are interested in the condition of large rigid body motion.5. Since the virtual work of the elastic forces does not depend on the rigid body states the corresponding components of the generalized force vector are zero.

the material is often not so well known for damping and so we will describe constants’ matrix one of the approximation techniques . Assuming that for two and we know the damping ratios and respectively vibration frequencies (in fractions of critical damping). This damping model is discussed in more details for the case of ﬁnite-element analysis in [1]. And so following the procedure outlined in the previous subsection we can compute the damping matrix ff similarly to the stiffness matrix ff . the Rayleigh coefﬁcients can be found from the equations: Similar to the previous section the resulting generalized force vector can be formulated as: f ê Note that given the and coefﬁcients one can establish the damping ratio that 25 1 A§ 3 % U & $ R ff f ¡ç ¥ç () ¡ç3 ¡ y ¥ç3 ¥ y 3 % ëª ê #" ê ª ê3 ¡ è 3 ¡¡¥ ª è ì 3 éèÁ U 1 m§ l $ 4§ l ² ² )) () w¬³µ³« é§ l "### ¡ ¡ © ¥ è ê U ç ”Mass proportional” damping pi ¡ © ç è ”Stiffness proportional” damping (2-38) (2-39) (2-40) .Rayleigh damping.Figure 2-1: Damping coefﬁcient as a function of frequency. However. The damping matrix in this case is calculated as: ff ff ff where (1/sec) and (sec) are constants to be determined from two given damping ratios that correspond to two unequal frequencies of vibration.

it might be reasonable to select some high frequency that limits the frequency band and set the corresponding damping ration to one thus damping out all the high frequencies. Since all the parameters used ( . 26 ê è ¥ ö÷¡ ç ®¡ ö ¡ ¥ç ¥ or. for which Rayleigh constants have been selected. Thus the system will become non-dissipative and generate vibrations during the simulation. Then. alternatively: g 3 ¥ ¥ ç & ¡ 3 ¡ ç f ¡¥ g 3 ¥ ¡ ç & ¡ 3 ¥ ç f ¡ 3 ¥ ¡¥ % H¥ 3 ¥ ç & ¡ 3 ¡ ç æ õ % H¥ 3 ¡ ç & ¡ 3 ¥ ç ä å õ ¥ y¡ õ ¡ ¡ç ¥ ¥ç ê ôè ñò ¡ © ó © í y © êîª è ¡ 3 ©ç y& y& ¡¡ ¡¡ øç ø ù ê ¤¤æ !è ¤ å ä¤ ©í ê (2-41) ç øç ê ê éè ð ¡ © 3 ï ç è è ê è (2-42) (2-43) (2-44) ç . Another use of Rayleigh damping might be to improve the numerical performance of the simulation by damping high frequency vibrations outside the frequency band that the user is interested in. suppose that only one damping ratio for the ﬁrst body eigen-frequency is known. the values of the coefﬁcients are often selected based on the experience with simulation of similar structures when the results of a simulation were ﬁtted to the same experimental results. The negative value of one of the coefﬁcients leads to the negative damping ratio for a range of ﬂexible body frequencies. That is the same and are used in the analysis of similar structures. including the material. The magnitude of the Rayleigh coefﬁcients is to large extent determined by the energy dissipation characteristics of the construction. and ) are positive the following conditions follow from Equation 2-42 if we assume : In practice.is speciﬁed at a value of : The general view of this curve is shown in Figure 2-1. Simple analysis of this curve tells us that damping coefﬁcient has its minimum value at the frequency and that the coefﬁcient grows linearly for higher and hyperbolically for lower frequencies. For example. . In fact. one of the features of Rayleigh damping is that the higher modes are considerably more damped than the lower modes. Having the two damping ratios and for two distinct frequencies the Equation 2-42 can be used to obtain the and coefﬁcients. The solution of the linear system Equation 2-39 gives the following relations for the and coefﬁcients: When choosing the damping ratios and frequencies it is important to make sure that the resulting and coefﬁcients are positive.

5. In a partitioned form. and the vector of generalized forces associated with the elastic generalized coordinates of the body.3 Generalized External Forces The virtual work of all external forces acting on a body can be written in compact form as (2-45) where is the vector of generalized external forces associated with the body generalized coordinates.4 External Point Force Let us assume that a force acts at point body. Let the position of the point be deﬁned by the vector (g) . (g) P/g (g) b/g (b) P/b (g) P/g (b) where P/b is the local position of point with respect to the body coordinate system. velocities. respectively. with the is translational and rotational coordinates of the selected body reference.5.2. Generalized forces may depend on the system’s generalized coordinates . and on time. the virtual work can be written as (g) (g) e f where and are the generalized forces associated. Later in this section in order to provide simpler and more compact equations the small angles and actual moments vector are used instead of generalized rotational coordinates (Euler parameters) and associated part of the generalized force . That is we will use: (g) (g) e f 2. (g) The virtual change P/g is then deﬁned as (g) P/g (g) b/g (b) P/b f (g) (b) P/b f 27 g lf § ª ¬£ £ § Ü 3 t 3 « § Ü 3 § e 3 ¬w« & § s Ü C ª ª ¬ £ § Ü 3 t 3 £w« § Ü 3 6 § e 3 ¬« 5 § C © ª « § e 3 ¬£w §e §s e © § g § e Ü T § r f é Ü 0))( 1 § Ü Ü § §s Ü §e "## $ 6 T g mT g T g §lf §lf § l rf5 m ¤g §lf gE@ § f §r §ú §r é Ü §s Ü §e Ü §e Ü § g (g) e 0))( 1 § ú§ Ü Ü §s Ü "## $ 6 T g T g T g §lf §lf § l rf5 § Ü T § l é Ü g §lf é Ü T g § l f §lf §e §e § §l (2-46) (2-47) (2-48) (2-49) (2-50) . The virtual work of this force is deﬁned as (g) (g) P/g (g) (g) (g) of the deformable where is the global position of point of the deformable body.

In terms of the generalized coordinates the virtual orientation change is then deﬁned as: (b) f f (g) (g) (g) f 28 § Ü ) () 1 § Ü Ü § §s Ü "## C $ 6 § C ¬wp @ ¬£w @ % 5 £« « § ª (g) r (b) f © is a virtual change of the the global orientation of the point where vector rules apply to the inﬁnitesimal rotations one can write: (g) © §e (g) gE@ §f ¬ § Ü 3 £w« g § Ü T § § § f § Ü é Ü § Ü § Ü § Ü § Ü g (g) e (g) (b) P/b (g) and were used to simplify the equations. § r 3 }«µ£¬ 3 T t ))0( 1 § ú§ Ü Ü §s Ü "## $ 6g 3 g t £w« f @ § e 3 ¬£w« & @ s 5 T § r f ¬ }«µ£¬ ¬£w«T T § e & î§ e m g l f @ r 3 }«µ£¬ 3 § e ng l f @ r § § § § © where the shape matrix as deﬁned at point .5 External Point Moment Let us assume that an external moment is applied at point of the deformable body. The virtual work of this moment is deﬁned as (g) (g) (g) where r is the virtual change of the reference frame orientation and f is the virtual change of the material particle orientation inside the ﬂexible body as deﬁned in Section 2.The vector (g) (g) P/g (b) P/b f (g) (g) (b) P/b f and so the generalized forces in Equation 2-47 can be recognized as where the facts that 2. Thus the virtual work é Ü ) () (g) 1 § Ü Ü § §s Ü "## $ 6 3 t w« @ § e 3 ¬« & @ s 5 ¬£ £ t §e Ü é Ü §lf §e Ü (g) P/g can be written in a partitioned form as (2-51) is deﬁned (2-52) (2-53) (2-54) . Let the position of the point be deﬁned by the vector (g) .6. Since (2-55) (b) (2-56) .5.

magnetic and electrostatic forces.5. where is the gravity acceleration constant vector and is the mass density of the body. Using the equations developed in Section 2. dimension N/m (Newton per cubic meter) is acting on each volume particle For the case of gravity this load is given as .6 External Volume Load In this subsection we will analyze the external loads acting on all of the particles of the body.3.4 and integrating over the volume we can get the components of the generalized body load : (2-59) (2-61) where and were deﬁned in Section 2. The most common cases of such loads are gravity.where f is a matrix calculated by differentiation of Equation 2-68 with f respect to the elastic states: (b) f f is the shape matrix deﬁned at point . For the case of the gravity force the equations can be further simpliﬁed considering that : 29 ² ü 3 }«µ£¬ 3 x t ` ³ w« § ² ü 3 ¬ 3 vt ³w« § }«µ£¢ § ` (2-60) (2-62) É §ü g § C § 3 }«µ£¬ 3 T § C f (b) f f g ²ôý 3 3 m ¨§ ²³ « ý § 3 }«µ£¬ 3 v x t ¤g ¨©l f § ³ w« § }«µ£¬ ô ý 3 H§ t g ©l f ² ©l f ¨§ ³ « § 3 xt É É ¦ Ç ² ü 3 }«µ£¬ 3 §T t ¤¥dÅ ² § ü 3 «}µ£¬ 3 T t ÇÈÅ ³w« § ¢£w« ³ É É ² ü 3 }«µ£¬¢ 3 ² § e ¡Ç Å ² § ü 3 }«µ£ 3 ² § e ¡Ç Å ³w« § ¢ µ³¬ £¢« ¬ µ³¬ ³ É É Ç g Ç 3 ² § ü ÈÅ ² § ü ² § ü ÈÅ þ°l f § ³w« ³« ³w« þ§ ÿl §ý © iC jC 1 C & C $ ktC itC y )) t C & k t C C ### ` )) j C # )) () t C & t C "### g lf @ § §ý 3 § 3 }«µ£¬ £ }«µ£¬ ¬w«T ng @ m §lf % § C C § §Ýp T 6 T t @ T t @ T 5t ' û § û § g (b) (2-57) vH§ t g °l f þ§ m ng §lf þ§ °l f t (g) (2-58) . 2. Finally the resulting generalized force can be recognized as: (g) e (g) where the fact that was used to simplify the equations.5. For generality we will assume that some load vector with the .

The internal stresses do not affect the elastic behavior of the body as long as the material is intact. That is a simulation tool needs to provide some means to deﬁne the geometry to be integrated. These distortions are especially signiﬁcant for large rings where they often cause manufacturing problems and therefore should be incorporated into the ﬂexible model [10]. One useful application of this approach is area load.Note that the derivation as presented in this section can easily be generalized for other kinds of loads that do not depend on the elastic states. One of the special effects that is of interest for this simulation is the internal stress release during grinding of rings. This approach is essentially a special case of the closed boundary speciﬁcation where the boundaries have simple conﬁguration. when the speciﬁed sub-volume is completely inside the body. Parametric speciﬁcation The sub-volume and sub-area can be speciﬁed by coordinate or parametric intervals. e. Both for the area and volume speciﬁcation two approaches are possible.7 Generalized Forces from the Internal Stress Release during Grinding The structural elasticity model presented in this thesis was incorporated into a grinding simulation tool [12]. 2.5. Following the procedure described in Section 2.. From the user interface design point of view.1 we can get: (2-63) where the volume integration is done over the material volume that is ground away during the simulation and stands for the residual stress tensor.g. Closed boundary A sub-volume can be deﬁned by a set of surfaces and a sub-area .by a set of curves on the surfaces. the parametric speciﬁcation simpliﬁes the integration procedures since the integration ranges become constant. When stressed material is removed the released stresses result in elastic distortion of the body. The internal residual stresses in the material can be caused by material phase changes in the hardening process. In both cases the boundary must be closed to make integration possible. When the material is removed they result in additional generalized force that affects the generalized elastic coordinates but does not inﬂuence the rigid body part of the system.5. where a constant load acts over a surface area. For simulation purposes it is often possible to assume that all the surface particles are subjected to the same treatment and only a thin layer of material is removed 30 « T @ « É t Þ ×§ Ç ÅÁ ©q § l § . That is as long as the load (force) vector does not depend on the point position within the body one can sum (or integrate) over the points once at the start up and use the precalculated shape integrals during the simulation. For some cases. the body and area loads rise an issue of sub-volume and sub-area speciﬁcation in the input.

or very hard. The values directly computed from the state variables. The main idea of this procedure is to delay the real force value for a small time 31 g k ~@ @ § f| ©q § l ©q § l g ¸@ w f Ég ©q g ¬ « tw £ f ³ ² Þ Ç Å T ³ ² § f ²´§ l ³ g ¸@ w ¢ f« ¬ £ g ¸@ w ¢ f« g ¸@ w ¢ f« « ~ ¬ £ .5.5. to evaluate the force function at every time instant. Some experiments done in the BEAST tool show that at least second time derivative’s continuity is needed in order to get reasonably large time steps in the solution. The keeps the value of the height integration in Equation 2-64 should obviously be performed over this surface mesh.8 2. The procedure is actually general and can be used for smoothing of any kind of force input that causes numerical problems. High order ODE solvers used for dynamic simulations assume high continuity of all the functions involved in the problem. In order to avoid this discontinuity problem an interpolation procedure was developed. Therefore the extrapolation procedures inside the solver can not predict the change of the resulting force. In a simulation the ground surface is described by a mesh. Every node of this mesh ground at the point on the surface. leading to much longer simulation times. from the ODE solver point of view the force changes discontinuously. Assuming that a rotation matrix transforms the body coordinates into the local surface coordinates one can perform integration of the shape matrix independently: (2-64) Note that for the case of ring grinding where a cylindrical coordinate system is used as the body coordinate system the matrix does not depend on the angle . In such cases some extrapolation or interpolation procedure must be used to provide a continuous force function in time domain. The interpolation procedure described here can be used for the forces which are changing with lower frequencies than some other fast processes in the simulated system. It means that the generalized force can also be evaluated only once per time step and can not be accurately evaluated between the are accumulated during the simulation run and can not be steps. That is why it is realistic to model the internal stress tensor as a constant tensor in the surface layer when the local coordinate system (s) normal to the surface is used to calculate the strains and stresses. During a grinding simulation the values are updated only once for every ODE solver time step and are kept constant when the solver does trial time steps during internal solver iterations. For a high-order solvers with adaptive time step size like CVODE force discontinuity results in convergence problems and shorter time steps.during grinding. The problem with the evaluation of the generalized force comes from the fact that the volume representing the ground material changes with time. Hence. The procedure is discussed in Section 2.8 An Interpolation Method for Forces with Discontinuities Some non-linear effects modeled in dynamic simulations make it impossible.

This value is delayed and which is seen as the expected force value for time . Before using the procedure the interpolation time delay parameter must be selected. at every time . Initial value of the force variable to be interpolated must be given. First and second time derivatives Delayed value of force and . assigned to the variable So. The only requirement on this parameter is to be larger than the maximum possible time step of the ODE solver. . A continuous force function generated for the discreet input . The only input for the block is the calculated force . . interval and to use cubic interpolation to get a smooth change of the force function between the current and the delayed values. Having the four parameters above (two values and two derivatives) it is possible to calculate the approximate value by using cubic interpolation: (2-65) 32 ª vÜ E gE f g ©E E g E E vÜ y vÜ & & f ¡ © & f ` ©± gE '© vÜ gE f is g©E © g©E © f± f g© E ± © A A f g©E f g©E f g©E g©E g©E f± fA f ª vÜ ©E Interpolation between and g ª © E ö E © E vÜ f g©E f g g & ¡ ©E & E f 'v ` Ü © ©E & E f ¡v ` Ü & ª g'g©E gE f © `f g©E & & E f 'v ` Ü © Ef© ª A & `f© E E gE f . The interpolation procedure can be seen as a ﬁlter block shown in Figure 2-2. Figure 2-2: Force interpolation as a ﬁlter block. The initial ﬁrst and second time derivatives of the variable are normally assumed to be zero if no other information is available. such that the interpolation block is using the following four variables: Force variable current approximating value ..

The values of the force variable for the time instances and are calculated using third order interpolation between and for and between and for . and using the procedure above. The corresponding value of the force variable is interpolated from . Note that is a continuous function which is the essential output of the interpolation block and should be used instead of the discreet values . .Figure 2-3: One dimensional force interpolation. In order to move to the next time step and to interpolate the values for times between and we need to evaluate and the derivatives and according to the equations: (2-66) Figure 2-3 shows an interpolation scenario where the ODE solver evaluates one extra time instance between two time step and . When the numeric ODE solver has done some iteration and decided the step the force variable value can be size and state variables values for the time calculated. 33 ¥ © g g © E v Ü © E & ` f © ± ª g © E ¡v Ü © E © A $& &¥ f &¥ f " g © E 'v Ü © E © " ª g © E 'v Ü © E © #& ¥ © ± &¥ f &¥ f " ª gE y vÜ E ª ©g E & v Ü¥ © E f & ` f © & ¥ © f © ± ¡ g © E ¡ & ¥ © E f & ` f © A g E 'v Ü E © ª g © E 'v Ü © E © & ¥ © A ¡ © & ¥ © f ¡ &¥ f ¥ ± © ¥ A © ¥ © ¡ © E ¥ © E E ª v Ü ¥ © E g©E f g © E ¥ f © ± ¥ © © A © © ¥ © E © E ¥ © E ª v Ü © E ¬ E ¥ © E ¥ © © ¥ © ¬ ¥ © E ¬ !E !E !E gE f © ©E © © .

x (b) f. For example if the simulated body is a part of a complex system and it is desirable to analyze the body movements from a certain known position in the system. 2. The angles can be calculated as: (b) f.y 2. However for the case of small deformations where only linear terms are taken into account the rotation can be analyzed as inﬁnitesimal. From the mathematical point of view. any element is changed in shape.7 Using Intermediate Coordinate System It is sometimes convenient.z iC jC 1 §e C & §e C $ kC iC C & § e C ### `y )) )) j § Ce k C ### § e C & § e C "### )) () g @ @ % % %f && % $ ª Ì % % && s £ % % "# % 1 %% $ () % "# (2-67) s (2-68) . In order to calculate the orientation of a material particle for the case of large deformations one has to use the properties of the deformation gradient [1]. The interpolation procedure can be applied element-wise to vector variables and hence can be used for generalized forces’ interpolation. this known position is a coordinate system with known motion.The interpolation procedure described in this section provides continuity of the second time derivative of the force vector which is sufﬁcient for most cases. In such a case a special care must be taken when working with time derivatives of 34 )) (b) f.x (b) f.y 1 % % % () && (b) f. and/or more efﬁcient. The rotation matrix that performs the rotation of a particle from the undeformed state to the inﬁnitesimally deformed orientation can be calculated as (see [21]): where is an identity matrix and are three small angles corresponding to the rotations around the three respective coordinate axes. for the numerical solver to observe the motion of a body in a coordinate system other than the inertial one. and rotated [24]. In order to analyze systems with rotational springs and dampers it is necessary to be able to calculate the orientation of a material point in the deformed body.z (b) f.6 Rotation of a Material Particle due to Deformation In general during the deformation of a body. It is certainly possible to achieve higher order continuity by using higher order interpolation methods. translated.

Hence the position of the body in the coordinate system is deﬁned as Then Equation 2-69 leads to the following expression for the linear velocity of the body: (2-73) Equation 2-71 gives the transformation for the linear acceleration: 35 g §s g 2 £ § s !4£ 2 § f 2£ §s 2£ ' '0 ' Q' !£ 27 f !£ 7 ª B 2 s !£ 27 y ª s !£ 27 ª 4 § 4 2 § A§ 4 § § 4 A § B 2 !£ 2 A § 4 2 £ § s !4£ 2 § ª B 2 2 £ A§ s ª B 4 !4£ 3A§ s B 4 !4£ A§ s 2 ª !4£ 2 § !4£ 2 § B 2 !4£ 2 A § B 4 !4£ 2 A § §s !4£ 2 A § 4!£ 2 § !£ 2 4 A§ 2 £ § s ª !4£ 26§ s !£ 2 ª B 2 s 4 § A§ ª B2 2 £ A§ s ª B 2 2 £ §± s !4£ s 8(' § ' 4!£ 2 y ª § ª B 4 !£ 2 § ± s B 4 !£ § ± s 4 4 B 4 A§ s §s B 4 4!£ 7A § 2 as rotation matrix . are given. That is. Below. System is a coordinate system with know motion in .position and orientation of a body. relative angular velocity are known. To be more speciﬁc let us assume that two coordinate systems are involved in the analysis. the transformation rules for taking time derivatives with respect to different coordinate systems. linear speed . the position vector . the following transformation is valid: (2-69) B 4 !£ 6§ ± s 4 2 )(' ' 4!£ 2 § B 4 !£ 5A§ s 4 2 4!£ 2 2 ' !4£ 3§ s 10' ' )(' B 2 §± s 2£ §s B 4 §± s System is the inertial coordinate system where the Newton-Euler equations are valid. and relative angular (2-70) (2-71) (2-72) (2-74) . and linear acceleration as well acceleration Then for a physical vector Applying Equation 2-69 on the relative angular velocity vector shows that the time derivative is invariant with respect to these two systems: Therefore the coordinate system where differentiation is done can be omitted from . the speciﬁcation the angular acceleration vector Using the same Equation 2-69 twice we arrive to the relation for the second time derivative: Now we will consider a body coordinate system deﬁned relative to the intermediate coordinate system . That is the position of the body is deﬁned with the vector and its orientation with a rotation matrix .

for the angular speed of the body and angular acceleration Note that the equations above do not specify the coordinate system where the components of the vectors are expressed. ) may be necessary to apply to bring all the vectors in the same coordinate system (e.. 2. The blocks in the ﬁrst row of this matrix ( can be computed analytically.g. and @Aûû §B §B g E k @ f and B 4 !£ § 4 § CC A§ § §B Ak 2 £ § !4£ 2 § ª B 2 2 £ § 2 £ § 4!£ § § C A § ²© 4l µ³¬ § § CC A§ ) () : C C QI P²© 3 § B & © § B HG3 E© F ² µ³¬ A § © C µ³¬ 4§ l C ¥ él ²© 3 © ²© µ³¬ § µ³¬ A § § 1 A§ § A§ § CC A§ § C C A§ C C C § C C A§ C $ C §§ C "## A ª ª !4£ !4£ 2 § 2 A§ B 4 !£ A § B 4 !£ § 4 4 !4£ § §B C ²© C µ³¬ A § (2-75) (2-76) 2 £ 9 @ !4£ 2 9 Ú § @ § Ù CD§ B § § CC A§ con) .8 Calculating Jacobian Jacobian computation is an important procedure that inﬂuences the efﬁcientcy of an that implicit solver. Let us rewrite the Newton-Euler equations presented in Section 2. Therefore appropriate transformation matrices (e.g. rotational and elastic coordinates. The ODE solver receives the differential equation represents the mechanical system behavior and it requires the partial derivatives in order to construct the Jacobian.2 for some body : (2-77) Differentiating both parts of the equation by the complete state vector we get: (2-78) The term involving the mass matrix derivatives introduces the speciﬁcs of ﬂexible body analysis where the mass matrix is not constant. 36 d § CC A§ where includes all the translational. Since for the elastic part of are zero and corresponding parts of are an identity matrix. the ”elastic” parts Calculation of the second row blocks ( and ) is more complicated and involves differentiation of the acceleration terms with respect to the state variables. Another property of the elastic case is the dependency of the generalized force vector on the elastic states. f f. knowing that it is convenient to view the partial derivatives as a set of four matrices: tains all the velocity terms. From the application point of view..Similarly. ’B’).

For this case in Equation 2-78 only the generalized force vector is dependent on and we can immediately write the differentiation chain: (2-79) where is the contribution to the generalized force 37 from the force acting on 7k yw û x §û and using force function only to compute a low-dimensional matrix §r W c bY ` d . Figure 2-4 illustrates the case of a force acting between points ﬁxed in the material of the elastic bodies and . Let us describe the procedure to calculate partial derivatives of the vector with respect to the states of the both bodies. This is done by using chain differentiation rules is the relative motion tensor. when (mixed sub-Jacobian) differs from the case . The procedure for calculating .y x y x y cG Figure 2-4: A simple force acting between two ﬂexible bodies. relative orientation and relative speeds of the points including the effects of structural deformation. For the application efﬁciency the most important aspect is to minimize the number of the force function calculation. Obviously the derivatives of the velocity states are nonzero only when differentiated with respect to the same body states or with respect to the states of a directly interacting body. that includes the distance between the points. In this context interaction means existence of a force acting between the bodies. where x W XSTR W Va Y ` c © §l à 7 §B C 7£© §l C à Ö d Uf hgW Y e 7© C 3 £© § l C A§ svsrp f utsrqi pY d © A§ @ § B û § 7 §B C © A§ C û U VbY ` ca Uc d Y` VTR US à d 7£© §l © £ ²7 § w ³© . We will ﬁrst describe the simpler algorithm that is used for mixed sub-Jacobians.

Continuing the expansion we get: The partial derivatives reﬂect both the motion caused by rigid body motion of the geometry center of body and the elastic deﬂection motion of the material point due to the changes of the elastic states. . from Section 2. is more complicated for the following two reasons: is a function of the state tensor Mass matrix Generalized force is a function of both applied force and the state tensor Recall. Note that from the differentiation of Equation 2-77 with respect to it follows that the matrix computed only once.body from body . Current implementation uses forward differences to calculate the partial derivatives on the right hand side of Equation 2-80. in order to get a possibly simple implementation we use the following general rule. the calculation of the derivatives .e. i.4 that the generalized external point force can be written as: (2-81) and so the partial derivatives for the elastic states become (all vectors and matrices are in the body geometric coordinate system): (2-82) Currently. Differentiation with respect to the same body state. Given some general function the partial derivatives with respect to can be calculated as: (2-83) 38 © §B kÜ g Ì k @ g Ì k f f & g k Ü ª Ì k @ k Ü 3 g Ì kk C C ª g Ì k f f f Y © §B C g Ì C 3 ©Y 1 © B C g 7 £ © § r ª )) Y § C © Y § © Y Ì § f )) 7 £ © § r áY © B C )) 7 £©§ § r ) () 7 §B C © £ 7 §w C ) () 1 7 © 7£© £ §r © §B gk@gk f f ©£7 3 ©² 7³ £ ²© ³© g § r 3 g © © Y Ì § f ª T © Ìt $ Y § Y §f 7£© §r "## © ¥E C C § w 7 £ ²© § r C 3 ³ © ²© C § r 7³ ©£ é§ l § f T t ª 7 © g© Ì £ §r Y § f §@ lû û § 7 §B C 7 £ ² § l C © ³© 7£© §l à t ### # "## ## $ § u xw û §B û © Y © §B C 7£© §l C kC gÌk@gÌk C f f à d @ §§ B û û k 7£© §l d is a constant matrix that needs to be (2-80) ..5. for instance.

q § § l B û g7 B@© B © 7 § § f £ §w we can construct a function for numerical differentiation that computes g© B@ 7 © 7 © Y § £ §rf £ §l pseudo code for the function follows: For each component of do: Choose an increment delta for forward differences Calculate: . compose kÜ . The u û § § rw û So. having computed and having functions 39 C k g Ü ó 7 £ Ì© § l & 7 £ © § l f 7 £ Y © © § B l C g © § B @ 7 £ © § r f 7 £ © § l 7 £ § © § l C 7 £ Ì © § r ª g © £ Ì 7 § w & © £ 7 § w f 3 © £ 7 § w C 7 £ © § r g 7 B @ © B 7 £ © © § 7 r © £ 7 § § kf Ü £ ª W k §W wk © § B W kH Y © § B and .

and there will be no induced vibrations 40 . We use the proposed set of functions to describe the deformation through the material. The use of the well-known analytical functions enables efﬁcient calculation of deformation shapes. The ﬂexible model that employs general shape functions is easy to use since no interaction with other tools are needed.Chapter 3 General Shape Functions 3. Thus. The series are known to be efﬁcient in approximating smooth functions and we can expect the real deformation shapes to be smooth and so accurately represented even with low order polynomials and small number of waves. Having considered the limitations with the reduction techniques we decided to use the classic approximation method and sets of general functions to approximate the deﬂection. it makes it possible to include effects of initial stresses and material removal. and the ﬂexibility is controlled by a few input parameters. Since we have a continuous description of the deformation ﬁeld. The integrals may be modiﬁed during the simulation.1 Motivation We observed that there is a large class of applications with bodies that are constrained by surfaces that can have a contact. An additional attractive feature of the approach is the ability to select speciﬁc elastic degrees of freedom in an easy way and disable the deformations that are known to be insigniﬁcant. we can have contact or connections anywhere on the body. The essence of these constraints is the need to have a continuous shape function description for deﬂection of each of such potential contact surface. The shape function integrals are automatically calculated at start up since we have a full geometric description of the body. Particularly we used Chebyshev polynomials to model deformation in radial and longitudinal directions and Fourier series for the circumference direction of rotating bodies. Since general functions are able to accurately approximate any set of eigenmodes this assertion also supports the idea of using general shapes. In [19] it was pointed out that the same deﬂection solution can be obtained using different sets of eigenmodes as mode shapes with high accuracy.

We believe that different complementing methods need to be available within a large general purpose MBS system. i. but those limits are yet to be found. there is a practical limit when this technique is not suitable. (3-1) . In the case of small area of potential contact for a body with complicated geometry one should complement the reduction method with the continuous shape function description for the contacting surface [30]. tangential or axial) is approximated as: e g Ì Ë e ©V 3 g ½ Ì Ë ½ d ¹ Â È3 g á Ì Ë ©V 3 e W i Ì © Ì © Ì © ª Y f Y f À Y f Y© ¥ © ¡ Y Y © e Û ¡ Û ¡ Û ¡ e e W Y i Ì XÛe ¡ © Ì dÛ ¡ © Ì XÛ ¡ © xw W Y ² § e © Y © ¡ Y © W µ³¬ 41 R Ì Ë e 3 g ½ Ì Ë ½ d ¼»S¹ 3 g á Ì Ë ©V 3 g © º Y f Y f Y f h For bodies with well deﬁned interfaces the mode shapes should be selected using eigenmode. In such cases standard reduction techniques will be more suitable. static and frequency response analysis as described in [18. Of course.e. 3.. when working in cylindrical coordinate system series of Chebyshev polynomials were used for the radial and axial directions and Fourier series for the circumferential direction. 29. The approach is based on the idea ﬁrst suggested in [30] for the representation of interface surface deformation shape only. 28]. In this work the same set of functions was used to represent deformation ﬁeld through the entire volume of elastic body. This is also the only approach known to work in simulations of prestressed structures during material removal.2 Choice of Shape Functions For the majority of real-life problems the exact deﬂection shape functions are not known and different methods to select shapes that provide the best approximation of the exact solution are often discussed in the literature. The approach taken in this work was to use series of general functions that are known to approximate any continuous function.due to modeling errors. for components with very little constrained. That is the deﬂection shape along certain coordinate direction (radial. For the case of multiple contacting surfaces and relatively simple geometry the general shape functions approach presented in this thesis should be used. complex geometry. It is also known that different sets of functions give approximately the same results when sufﬁcient number of degrees of freedom is included in the system. Speciﬁcally.

That means the only condition imposed on the shape ﬁeld is the absence of rigid body motion degrees of freedom in the elastic deformation [18]. In order to understand the importance of this condition consider a small example in 2D. These derivatives are required for the strain and stress calculation.all the bodies have all the degrees of freedom. If we follow the standard procedure for calculating the strain energy of an object we will get: where is the Lame material constant and .Where give the number of functions used in each direction. The relation means that when 42 ² i ² i ¤¡ & ¥ ³ ³ Ë ² i g j @ i f w ³ ¥ 7 h e g f WY g R Ì Ë e ©V 3 h Ì Ë g V 3 V Ì Ë f ©V 3 e g W f i Ì © Ì g © Ì f © g © g Y f Y f Y f Y© © Y Y © e Û ¡ dÛ ¡ dÛ ¡ k gi E gi E iy gi & f¥ f¡ i g if g i f ¥Ì ¤æ å f ä¤ gi ª i 9 ¡ ¡ ¥ f g É É É g i i E É & g g i i E É i ª g i f ¥ E f y g i i É É f¡ f¥ fi ` g i É É ¤¤¤¤æ å g i f i ¥Ì É ä¤¤¤¤ f § G§h f ` à % Ë ² i ³ ¤¡ @ WY W xw j i ½ WY @ WY @ WY á 9 @W Y W ² e µ³Y ¬ § h (3-2) (3-3) (3-4) g j@ i w f . or ) is approximated as: where give the number of polynomials used in each direction. for the Chebyshev polynomials the relation is given below: Note that having the recurrent relation for the polynomials we can easily get recurrent relations for the derivatives of the polynomials as well. For the case of Cartesian coordinates Chebyshev polynomials were used for all the three coordinate directions.area of the body. In this thesis the case with no ﬁxed external boundary conditions is considered . Thus we have to impose some extra conditions in order to make the selected set of mode shapes admissible for the problem. Hence the real deﬂection shape along certain coordinate direction ( . This is not true for the general mathematical functions used in this work. is the Chebyshev polynomial of order and is the vector of elastic coordinates. An important property of the both orthogonal polynomials and Fourier series is the availability of the recurrent relations providing a way for fast series calculation.3 Separation of Elastic and Rigid Body Motion Modes Typically in an assumed mode shape method the selected mode shapes always satisfy some speciﬁc boundary conditions. For instance. Again using the Chebyshev polynomials as an example: 3. Suppose that only two deﬂection functions are selected: and .

In Figure 3-1 we show the changes to a simple square when the shape coefﬁcients are . since the change of the area doesn’t create any energy and consequently doesn’t provide any stiffness. The conditions that were chosen to remove the rigid body motion terms from the deﬂection functions correspond to elastic body mean-axis condition for the reference frame. The example shows that the use of general functions without proper condition will not just slow down the simulation. The stronger version of the mean-axis conditions can be formulated in term of the precomputed integrals (see Section 2. The mass matrix simpliﬁes to (using body coordinate system for all vectors): ff 43 6 h` & @ ` 5 !G h§ d E j 1 Q | e°¢ ¯ $ x t x t () x t % { "# % g ¥ g ¡ x t § 3 1 g ' ¥ x t () ¡ ' x t % § 3 x t ª ¥ t § (3-5) & ¡¥ x t f & ¥'xtf $ & ' ¡ x t f "# (3-6) (3-7) .no strain energy is created. 26]. The mean-axis conditions are obtained by minimizing the kinetic energy with respect to an observer stationed on the ﬂexible body [18. It is obvious that the area of the square grows.3). for instance. 19. would lead to the ’explosion’ of a body. The immediate conclusion that this shape combination corresponds to the rigid body motion is only partially correct. The total of six conditions can be partitioned into the three conditions for the translational degrees of freedom: And three conditions for the rotational degrees of freedom: Note that the conditions effectively attach the reference frame to the mass centre of the body allowing simpliﬁcations to the dynamic equations described in Chapter 2. x x 2 B 1 A y 1 C D 2 B A 2 1 y 1 C 2D Figure 3-1: Use of shape functions without boundary conditions. The discussed example. but can result in some non-physical effects. for our example it is doubled.

Having the mass matrix simpliﬁed and noting that from the no mass centre motion condition one can also avoid the inverse of the complete mass matrix when f ﬁnding the acceleration tensor from the Newton-Euler equation and get: (b) m/g (b) f Even though the expressions in Equation 3-8 might look more complicated than the original equation in Section 2. Let us recall the deﬂection to strain equations for the cylindrical coordinates in Equation 2-30: In order for the strain components . that is for .2 one should note that the simpliﬁed expression require inverse only of two matrices: a constant matrix ff and a small matrix to ﬁnd the angular acceleration.And the quadratic velocity term also becomes zero. The problem here is the singular point in the centre of the coordinate system. but it is unavoidable for the solid bodies (solid shafts). which is quite natural for the bodies with rotation symmetry. Hence the time for the solution of this linear system grows instead of in the original case with full mass matrix. the following equation must be satisﬁed for any g § s 3 T x t & èd3 T x t & § l f 3 ¥ E m ± g gwg § ± s 3 T x t & § l 3 f ¥ E 3 yx t & § l f ¥ E T x t 3 ¥ E 3 x t & x t f §l ` r (b) ff ff f m/g (b) ff f m/g 44 ÷ ' ¡ ß ¡¥ ß n¡ ß ¡ ½ k C ª ~ CC | wC w ` | w C C ª xká w C C ½ | C ª g ½ w ~ CC | & á¨w f ` wC kw CC C g áxw ª ½ ~ w C f `| C x|á w C i g op§ l f % ¡n¡ ß ¥ n¥ ß '¡ß '¥ß ¡¥ ß ' n' ß | g j ' f % §± 3 x t g j ¡ f §± § §± s (3-8) (3-9) . as 3. and to be ﬁnite in the centre of the coordinate system. This point is not important for the bodies with no material at the centre (rings and hollow shafts).4 Special Shape Functions for Solid Bodies in Cylindrical Coordinates One problem with the general shape functions concerns the use of cylindrical coordinate system as a body coordinate system.

21]. They are: elimination method. the cylindrical shape functions were modiﬁed so that they never result in a deﬂection of the centre line of the body. Lagrange multipliers method is very popular in multi-body simulation tools [1. The Cartesian shapes were then transonly a single polynomial formed into the cylindrical coordinates by using ordinal coordinate transformation: 3. However it is in general very hard or impossible to determine which degrees of freedom can be eliminated in a large simulation model. At the same time the Cartesian shape functions were restricted to the centre line deﬂections by removing the dependency of the and coordinates thus leaving in the series. The three most popular methods are mentioned in this section. This is done by removing the constant term from the Chebyshev polynomials that have as an argument. These extra state variables are called Lagrange multipliers. The elimination method is based on the idea that the set of boundary conditions can be used as a set of equations to express the values of some elastic state variables as functions of other states of the body. In this method one extra state variable is associated with every boundary condition.From the physical point of view the Equation 3-10 require the continuity of the displacement functions for every point on the centre line of the body. It provides a general way to deal with boundary conditions. The cylindrical shape functions were used to describe the deﬂection shape of the cross-section and the Cartesian shapes were dedicated to the central line deﬂection. That is. That is why this approach is in practice limited to the simplest cases when a boundary condition explicitly speciﬁes the value of some of the state variables. In this case corresponding degrees of freedom can be considered as redundant and may be safely removed from the system. A Lagrange multiplier becomes nonzero if the boundary condition is active and results in a reaction force that forces 45 áY Ì Ë ½ w º% 1 w $ 1 ` ¹% Â h ¼»S¹ $ 1 w $ w 3 % ~ ¼»ºw¹À ~ ~ ¹ hÀ & á Â "# () ¨w "# () w "# () % ~ gz Ì Ë Y f j i % % (3-10) (3-11) k angle % ½ w & á ¨w ª ~w CC ~C á¨w C ½~ C wC 6 nyz@ % 5 l~ m k and for any cross-section cut : . Lagrange multipliers method and penalty method.5 Imposing Constraints for Rigid Body Motion There are several possible approaches to impose boundary conditions discussed earlier in this chapter on the shape functions. In order to realize this constrain both sets of shape functions deﬁned in Equations 3-1 and 3-2 were used.

3. 3]. according to [3] is the sensitivity of the results to the computational errors. In reality. The modiﬁed stiffness matrix is given by: ff ff (3-13) (3-14) and an extra reaction force vector can be computed as: The large boundary condition spring constant is typically chosen to be several orders of magnitude higher than the maximum diagonal element of the original stiffness matrix ff . To avoid this problem the constant should not be set to extremely high values and double precision arithmetics must be used. In out test runs it was noted that the numerical error introduced by the penalty constants is several order of magnitude lower than the general modeling error and other kinds of computational errors. By choosing the parameter large enough it is possible to get sufﬁciently accurate solution. In this case boundary constraints introduce frequencies that are much higher than the body’s eigenfrequencies and are in practice completely damped out when even a small stiffness proportional Rayleigh damping is used.the condition to be satisﬁed. The idea of a penalty approach is to represent each boundary condition as a spring with very high stiffness [1. a solution where the error in boundary condition lies within the numerical tolerance. If the number of constraints is large (comparable to the number of the elastic shape functions) the number of extra state variables becomes comparable to the total number of elastic degrees of freedom. The stiffness matrix of the elastic body and the force vector are modiﬁed to incorporate the effect of this spring. The drawback of this. That is why it was chosen for the implementation in the BEAST system.6 Reducing the Number of Flexible States The series deﬁned by the Equations 3-1 and 3-2 provide a complete mathematical set of functions suitable for approximation of any continuous deformation shape. constraints often restrict only a few variables and so the vector is sparse. The penalty approach is general and simple to implement. However in a real simulation only some speciﬁc deformation shapes determined by the 46 ¬ o § p 3 Ì PÈ3 ¬ o ª T § p 3 ´§ p 3 ¬ ÌP T § 3 ´§ p o ¬ l § x ¬ o §p ¬ o §p . The drawback of this method is the necessity to have extra state variables and keep larger mass matrix and force vector. The detailed description of the approach is given below. It must be noted that the introduction of multiple constraints with the penalty approach results in large diagonal elements of the stiffness matrix. i. That is the size of the system of ODE becomes signiﬁcantly larger and makes the simulation slower.. A general linearized boundary condition can be written as: f (3-12) where elements of are known constants.e.

two complete revolutions of the workpiece in a grinding simulation is an example of such a repetition. eigenmode.ﬂexible body geometry and boundary conditions are present. Having selected the state vectors into the characteristic set and for a speciﬁed relative threshold (which is an input parameter) the ﬁltering algorithm given below should be applied. The characteristic set can be constructed using static. Trial simulation In a rotating machinery simulations there are often some periodic processes where the difference between the two periods is relatively small.. or quasi-static analysis (see Chapter 5). In such a case just the speciﬁed number of eigenvectors should be included into the characteristic set. the following revolutions of the machine can be simulated faster with smaller number of elastic states. Note that this approach is fundamentally different from the traditional reduction techniques. For instance. By detecting such shape functions and eliminating the state variables associated with them we can reduce the number of ﬂexible states for a body and speed up the simulation.g. Then. instead we just eliminate some of the shape functions in the complete series. Note that the vectors generated during the simulation often correspond to the same shape with different amplitude. In such cases a trial simulation of exactly one revolution can be run and all the state vectors during this trial simulation can be recorded into the characteristic set. We do not try to ﬁnd the most suitable shapes and change the functional basis. or using a relatively short trial simulation in the following ways: Static shapes The state vectors corresponding to the static solutions of a typical boundary conditions problem should be a part of the characteristic set. Quasi-static shapes In rolling-bearing applications simulation for a certain speciﬁed angular velocity is common task. Let us deﬁne a characteristic set of shape vectors to be the set elastic state vectors corresponding to the deformation shapes typical for the speciﬁc simulation. That is why contributions of some of the general shape functions to the particular solution are small. The solution vectors for the corresponding quasi-static case should be included into the characteristic set. after the reduction. Eigenvectors If the user speciﬁes a frequency band then the eigenvectors corresponding to the free body eigenfrequencies in this band should be included into the characteristic set. Mark all the elastic states as inactive For each state vector in the characteristic set q u B r© s Y© xi B ¥ ¡ Û ` t3 ß ri q – Calculate the threshold: ß (3-15) 47 . Alternatively a user might want to specify the number of eigenmodes (e. specify that only ﬁrst three eigenmodes are of interest).

The proﬁle curve is deﬁned piecewise and each segment has a separate parameter going along the curve. Hence the i-th element should be removed from the elastic state vector . Assuming that the contributions of the states with inactive mark are negligible they can be safely eliminated from the problem. The technique has not yet been used in the production runs. Cylindrical system for such bodies. As the result of the algorithm only the states that have very small contributions to the vector in the characteristic set retain the inactive mark. Integration of a rotating proﬁle body. damping and inertia properties of elastic objects are needed for simulation.7 Volume Integration Volume integrals representing the stiffness. and . The integrals are calculated in the cylindrical space and not in the proﬁle curve parameters’ space. most work has been done on integration of bodies deﬁned with a rotating is said to be the ’natural’ coordinate system proﬁle curve. if one percent error is acceptable.– Remove the inactive mark for the elastic states that have the absolute value larger than . in Cartesian coordinates gives low accuracy and 48 d ß g ~ wg z@ w k @ g z@ w f ~ f| f v W x t § ± xt ó y & È` e g k z@ @~ }|f qri § A vW x t ~ w ß z U z z §l § . To deﬁne rotating proﬁle body a proﬁle curve must be given for any angle . then can be set to . However some of our test runs showed that up to the half of the elastic states generated by the general functions from Equations 3-1 and 3-2 for the user-speciﬁed number of degrees of freedom can be eliminated later on by using the presented reduction approach. acceleration vector as well as from the generalized forces vector and precomputed integrals and . Here. . We assume that the curve is continuous both in value and derivatives within each segment. The choice of the threshold is normally determined by the anticipated modeling error. 3. since the deformation shape functions that we integrate are deﬁned in the physical space and are not related to the boundary parameters’ spaces. The integration of rotating proﬁle bodies has to be performed in cylindrical coordinates for accuracy reasons. velocity vector . Note that it is assumed that all the shape functions are scaled with the real geometrical dimensions of the body and therefore all the state variables has the same dimension (meters) and can be directly compared. That is. Elimination of an elastic state means removal of the i-th element from all the vectors that have a part associated with the elastic states. that often has some kind of rotation symmetry. Besides the i-th row and i-th column should be removed from all the matrices associated with the elastic states: . This is due to the fact that the general interpolation series introduce many shapes that correspond to very high frequencies for the particular body. Those states have therefore become the candidates for the reduction.

First. Secondly. We start by going through all the segments of the proﬁle curve and evaluating the boundary for the speciﬁed . The number of points must be more than the double of the number of waves used in the shape functions or in geometry deviation in circumferential direction. Select the number of integration points in circumferential direction. asymmetric stiffness matrix which results in non-physical vibrations in simulation output.7 to illustrate the algorithm. Integra. Integration is performed ﬁrst in the radial direction. then in longitudinal direction and ﬁnally in circumferential direction. In Figure 3. tion bounds are 2. The complete algorithm follows. we want to put the most varying boundary (radial) into the innermost integral to simplify the implementation and be able to use high precision integration scheme for it. The ﬁgure shows a six segments proﬁle curve with four straight and two curved segments. Our purpose is to ﬁnd minimum and maximum coordinate for each segment as well as the edge values between the segments. For each angle we have a cylindrical sector to integrate. We use extended Simpson’s (trapezoidal) integration scheme in circumferential direction since it is known to give good results for periodic functions and we have Fourier series as a basic set of functions in this direction.7 we 49 k v 'k k ¡k 6 ôm & 5 m@ ¥k ~ ~ Ìk . We will refer to Figure 3. There are two basic ideas behind this ordering.Figure 3-2: A proﬁle curve used to deﬁne a rotation symmetry body. Both in radial and longitudinal direction we use modiﬁed Clenshaw-Curtis quadratures [15]. These values are used to deﬁne integration intervals in direction. having integral in circumferential direction as outermost ensures the rotational symmetry properties of the resulting matrices. 1.

Every pair of radii in the lists deﬁne an integration interval in radial direction and quadrature formula can be applied again to get the radial abscissas and weights. and .7. . Note that each integration interval is treated separately. An ordered list of radii is build for each abscissa. Note that since we have split the integration along z axis in intervals the number of integration intervals in radial direction will be constant within each z interval. In the example. Figure 3. calculate abscissas and weights according to the quadrature formula.would ﬁnd 3 integration intervals with bounds . For each integration interval in direction. How does the complexity of the boundary geometry affects the accuracy of the integrals? How do we select the optimal number of integration points? Is adaptive strategy necessary? 50 ¡k Ìk 'k ¡k ¥k Ìk k 'k ¡k k . 3. We need to go through all the segments once again to ﬁnd the radii for each abscissa detected in the previous step. 4. There are still some uncertainties regarding the integration process. If the closed volume condition is satisﬁed then the number of elements in every radii list will be even for hollow bodies and odd for bodies with material in the centre of a cross-section. The order of integration (number of points) is double the highest order of polynomial used as a deformation shape in longitudinal direction. we have just one radial integration interval between and and two intervals between and Having computed the set of integration points and weights as described above we can approximate all the required integrals with the weighted sums of function values computed at the speciﬁed points multiplied with the corresponding weights. In the general case we might get several non-adjacent integration intervals.

**Chapter 4 Modeling Ideal Connections with Control Points
**

4.1 Control Points and Flexible Bodies

The BEAST system design uses continuous deﬁnitions of the complete surface geometry. No explicit interface nodes are predeﬁned. This is in contrast with most other tools where some joint attachment points are included in the basic model design or FEM-based tools where joints and force elements can be connected only to the ﬁnite element mesh nodes. However ideal joints and force elements are often used in multi-body modeling. The older version the rigid-body subsystem of BEAST allowed such forces to be introduced in the geometrical centres of the bodies or in a single control point between the bodies. The user of the system had to translate the combination of forces and moments acting between two bodies to the bodies geometry centres or the single control point. Since in the rigid body dynamics the force and moment couple applied at one point of a body can easily be translated to any other point (including the geometry centre of the body), this feature provided enough modeling freedom until the introduction of ﬂexible components. For a ﬂexible body the relation between the generalized force tensor and applied force is more complicated (see Section 2.5). That is why we cannot require the user to translate the force to the body’s geometry centre or any speciﬁc coordinate system by hand. Instead, there must be a method to specify any material particle inside the body volume as the point where the force vector is applied. The control point design discussed in this section resolves this problem. Control points provide a way to specify a particle’s position, orientation and speed inside a body (relative body geometrical centre), or in the model reference frame (called cB coordinate system in the BEAST framework). Thus a control point can be seen as an information data structure consisting of a position vector, orthogonal rotation matrix as well as linear and angular velocity vectors of the particle. These points can be used to study the elastic deformation of the speciﬁed material particle, 51

cg

Figure 4-1: Control points. and to apply forces at the speciﬁed locations (not only at the geometry centre). First, the parent system of a control point must be selected. This is the coordinate system where all the output components of the control point are calculated. Generally, it can be any abstract coordinate frame. The two most common cases are ﬂoating model reference frame and a body geometry centre. For each control point the following relations are deﬁned:

The position of the control point which is normally written in the output ﬁle. Every component of this relation should be understood as a combinations of position vector, rotation matrix and speed vectors. The plus operator represents a complete coordinate transformation (translation, rotation and velocity). The tensor provides a common interface to all the different types of control points described below. Parent coordinate system (model reference frame or geometry centre of a body) gives the coordinate system where the output is calculated. deﬁnes the beginning (base coordinate system) for the tensor introduced by this control point. tensor itself can be a control point making it possible to build ’chains’ of control points. Such chains can be efﬁciently used to specify complex motion as shown in the example section of this chapter.

is the relative motion deﬁned by the control point. The speciﬁc of this vector calculation determines the type of the control point (see below).

The example Figure 4-1 shows two control points deﬁned in a body. Both control points have body geometry centre as the parent coordinate system. The ﬁrst control 52

¬£ v v § e

** ¥ µ ¬¥ v v § e « £¥ v v § e £
**

¡ v á £w£ µ ¬ § e ¡ v v³ á¡ á £w£ v w v ² § e ¡ v á £w£ v v § e

c ` ¤E Q

¡ µ ¬¤¡ v v § e £

¡ ª ¡ µ £ v v § e v á £w£ ¬ § e v á £w£ v v § e ¬

c y ¤E Q

ª £ £ ¡ µ ¤¡ v v § e « ¥ v v § e « ¤¡ v v § e ¬£

¡ v á £w£ ¬ § e µ £ v v § e ¬

point is deﬁned relative to the body coordinate centre and so the output for it is equivalent to the . The second control point is deﬁned relative to the ﬁrst control point thus forming a ’chains’ of two points. Currently three types of control points are available in BEAST. They are: tensor for this kind of control point is constant Fixed control point. The and so the velocity components are zero. It is semantically identical to a time dependent control point (see below) with all the time dependent parts set to zero. Compared to the time dependent control point the ﬁxed control point is faster to process, requires less input and memory. Time dependent control point. The components of the tensor are functions of time. Note that it is impossible to specify both velocity and position as independent time functions and so user must either prescribe the position - letting the program to calculate velocity or provide the velocity and let the program compute the trajectory. Flexible control point. This type of control points can be used only within ﬂexible bodies. They provide a mean to examine the deﬂection of a speciﬁc material particle. Unlike the two other types of control points the ﬂexible control point does not have any input and it must be the last control point in a chain. That is it is not allowed to deﬁne other control points having the ﬂexible control vector is computed by BEAST and gives the point as a base. The deﬂection vector and deﬂecting speed and the output tensor speciﬁes the deﬂected position of the material point that had coordinates in the undeformed body. In the current implementation we do not calculate the orientation of the deﬂected point. Flexible control point deﬁne extra output tensors: and . They are the second Piola-Kirchhoff stress and Green-Lagrange strain tensors discussed earlier in Section 2.5.1.

**4.2 Boundary and Loading Conditions
**

Stiffness-dampers, force-moment, and simple parametric bearings can be used to set the boundary and loading conditions for ﬂexible bodies. The required input is identical to the one needed to set the conditions for rigid bodies. Special attention however must be paid to the type of control point where the force is applied. The reason for that is the different deﬂections caused by the same force applied to different parts of a ﬂexible body. If some loading or boundary condition should cause deﬂections then it needs to be applied at a ﬂexible control point in the body. If some speciﬁc loading should only inﬂuence rigid body motion then it should be applied at a time dependent or ﬁxed control point. 53

¡ v ¡ á £w£ ¬ § e v á £w£ v v § e

£ v v § e ¬

c y ¤E Q

£ v v § e ¬

dde | E ¯ ¯¯

£ v v § e ¬

¥ ¥ v v § e ¬£

}d ¦ | E ¯

c ` ¤E Q

The ﬂexible control points that are based on the respective ﬁxed control points follow the deformation of the ring.4. Four of them have the model reference frame ’cB’ as parent and four others are deﬁned in the ring. This connection can be of different nature. point is a time dependent control point that deﬁnes a reference frame The with constant translation relative to the cB system and rotation with constant angular velocity .3 An Example of Control Points Usage Figure 4-2 shows an example model that uses control points to apply complicated loading conditions on a ﬂexible ring. 54 c ` ¤E ÔQa P c ` ¤E bQa 0 Q ¡ § yxc¤E bQa 0 Q ¤cE ÔQa 0 Q c y ¤E ÔQa 0 ¤Qc E 0 ` bQa Q ¥ § c y ¤E ÔQa P c y ¤E Qba 0 Q ¡ § c ` ¤E Qba P ¥ § ` c¤E ÔQa 0 Q c y ¤E Qba P . Eight control points are used in the model. however. It is shown in Figure 4-2 that ﬂexible control points are connected to the control points in cB system. that some experience is required from the user in order to understand all the features of the control points and apply them correctly and efﬁciently. has as the base system. The structure of the control points’ chains used inside the ring is simple. It must be noted. . It is also a time dependent control point with constant translation and constant velocity rotation given by . Thus all the control points that follow in the chain will rotate around this control point position. The discussed example shows how a complicated loading condition can be described with the control points. Having follow the satellite-like motion as speciﬁed by the velocities arranged this motion it is now necessary to relate the control points to the material particles of the ring. Most often a combination of springs and dampers is used to prescribe a body motion with certain stiffness and damping coefﬁcients. Just two and ) and two ﬂexible control points ( ﬁxed control points ( and ) are used. The purpose of this model is to force satellitelike motion of the ring and deform it elliptically in the same time. Both and are ﬁxed control points having base system in . The next control point. The ﬁxed control points deﬁne the positions of the material particles inside the ring in the undeformed state. In both cases chains of control points are used. Since they have the same base system an imaginary line segment between the two points will not change the length but will and .

ctl4 55 ¥ § .ctlf1 bR.ctl1 cB.ctl1 cB Figure 4-2: Specifying loading conditions with control points.cB.cg bR bR.ctl2 bR.ctl3 bR. Object-oriented notation is used.ctlf2 cB. The coordinate system cB and body bR are the parent objects and the control points and geometry centre ’cg’ are the child objects. ¡ § cB.ctl2 bR.

None of them deals with the rigid body motion and mass. Eigenmode and Quasi-static Single Body Analysis Before running a complete multi-body dynamic simulation with ﬂexible bodies it is often useful to perform some simple analysis on a single body model. 56 . This analysis can help answering the question of the required number of deformation shapes for the particular big simulation case. Since the loads are static no inertia forces are active in the equilibrium state. often resulting in eigenfrequencies much higher than anticipated. For instance. Eigenmode analysis. They are: Static loading. The discussion of this case later in this chapter shows that only the properties of the stiffness matrix and deformation ﬁeld can be studied with static loading. Note that in order to perform static analysis the applied loads must be balanced so that no rigid body acceleration is necessary to resolve them. This happens since lack of shapes reduces the number of degrees of freedom for the material particles of the body. Three kinds of single body analysis were implemented in the BEAST system. specifying no (or too few) shape functions in a certain coordinate direction results in implicit ’ﬁxed-ends’ boundary condition.Chapter 5 Static. Free body eigenfrequencies and eigenmodes are computed in this analysis mode. In case of general shape functions some additional boundary conditions can be implicitly introduced by the selection of the shapes. Note that the presented analysis types are only meaningful for ﬂexible bodies models. Eigenmode analysis helps in studying the properties of both mass and stiffness matrices of the body as well as the properties of the selected deformation ﬁeld but does not give any information about the centrifugal forces acting on the body. In this analysis some constant (time independent) loads are applied on the body and body’s deformation in the equilibrium is studied.

It must be clearly understood that no time domain integration is required to perform single body analysis. That makes the analysis much less computationally demanding than complete simulations. Hence, users are strongly recommended to carefully study the results of single body analysis before performing the computationally heavy time domain runs. The following sections discuss the equations necessary to perform the all three kind of analysis described above. They all are based on the generalized NewtonEuler equation that was discussed in Section 2.2. The equation with all the vectors and matrices computed in body’s local coordinate system is repeated below for convenience:

(b) b/g (b) f

ff

**5.1 Static Loading Cases
**

The equilibrium equation for the static analysis can be derived from Equation 5-1 by setting all the accelerations and velocities to zero, and further assuming that no rigid-body motion is required to resolve the loading case. The resulting equation follows: (5-2) f ff f This is a linear equation for the elastic state vector that has a unique solution:

f ff f

The generalized external force vector f gives the speciﬁc of the loading case and normally stands for superposition of several point forces and body loads. See Sections 2.5.4 and 2.5.6 for the calculation procedure for these cases. Typical example static load cases for rings include symmetrical -point loads and for beams they are point or distributed loads acting on a simply supported beam.

5.2 Eigenmode Analysis

The eigenmode analysis discussed in this section is limited to the discussion of the free-vibration equilibrium analysis with damping neglected. More discussions on eigenmode analysis can be found in [1, 5]. 57

g g m mo 1 ¤g p§ l f $ ª 1 A § 3 U & ¤g§ 3 & § l f $ o pl f g §lf ) () g o§§ l f "## ) () §lf "##

§ 3

g §lf g § l f 3 ¥ z § E g & §lf %

) ()

e Q d | E °¢ j ¯ $ 1 §± § $ 1 x t x t § ± s "## ) () x t vT § t "##

Quasi-static analysis. Generally this kind of analysis assumes constant velocity analysis. The most interesting cases are constant angular velocity revolutions around different rotations axes. This analysis helps the study of the centrifugal forces as well as mass matrix, stiffness matrix and deformation ﬁeld of a body.

(b) R e

R

f

ff

f

ff

ff

f

(5-1)

(5-3)

Since rigid body frequencies (which are always zero for a free body) are not interesting to analyze, Equation 5-1 reduces to:

ff f ff

In vibration analysis it is assumed that the solution of this equation is represented in the form: (5-5) f where is a vector with the number of elements equal to the number of degrees of freedom (or number of elastic coordinates) f used in the system, is the time variable and is a constant identiﬁed to represent the frequency of vibration (radians/second) of the vector . In most cases the vector is called eigenmode and the associated constant is called eigenfrequency. Substituting Equation 5-5 into Equation 5-4 a generalized eigenproblem is obtained, from which and must be determined: (5-6) ff ff

ff

The vector is called the -th mode shape vector, and is the corresponding vibration frequency. In order to calculate the positions of material particles for a certain frequency one just need to substitute the vector ( is an arbitrary real constant) for f in deﬂection calculations (Equations 2-2). A typical eigenmode vibration visualisation procedure should repeat this calculations for different values of obtaining an eigenvibration animation sequence. The numerical algorithms for the solution of eigenproblems have attracted large attention in numerical analysis. Most numerical packages include these algorithms. In case only algorithms for the solution of standard eigenvalue problem are included in the package it is possible to transform the generalized eigenproblem into the standard form (see [1] for the discussion).

**5.3 Quasi-static Analysis
**

The quasi-static equilibrium equation is derived from Equation 5-1 by setting the

(b)

acceleration vector, the velocity vectors b/g and f , and the external force vectors to zero. Then for a given, constant angular velocity :

ff f

f

58

9

© í

9 © §¨ 3 9

g

A§

o ¢§ l f

ª 3 §

A§ s

¥ %

and

7 §¨ 3

&

3 ©T § ¨

%

The eigenproblem yields the orthonormalized, i.e.,:

f

eigensolutions

, where eigenvectors are M(5-7)

f

E

§¨

à à g©

¡ rØa¾aa ¡ Ö d@% d @ ` æå ä § ¨ @ ¡© f §¨ 3 ¡

§ 3 §¨

º E ¾»w¹ § ¨

&

!§ ¨ 3

±§ 3

§

§¨ d © §¨ §¨ §

f

(5-4)

(5-8)

(5-9)

This linear equation can be used to get approximate solution for the static case. Such solution is quite accurate for the cases with weak coupling between the elastic deformation and quadratic velocity vector, and can be considered satisfactory for most cases. To obtain the exact solution it is necessary to recall Equation 2-25:

**Substituting Equation 5-10 into Equation 5-9 and simplifying, we come to a linear equation:
**

ff f

If the equation above is solvable then the solution vector gives the deformation shape for the analyzed quasi-static situation. Absence of the solution indicates coupling between the speciﬁed rotation and some of the elastic states’ time derivatives f . In such a case no unique general solution exists to the quasi-static problem.

59

A§

** ª g g ¥ w7 ¥ r© ¥ w7 ¥ r© A § 70© Ð 70© ' Û ' Û 3 y & § 3 07© x t T07© x t f 07¡© f ' Û ' Û & Ò
**

f

S7 r© T07© x t 87¡© g f ¥ Û ¥ Û & ' '

g g ¥ w7 ¥ © ª § 3 07© x t 07¡© f Û Û ' '

f

g opl f §

f

(5-10)

(5-11)

1 Long Beams Distributed gravity load acts on a simply supported beam (see Figure 6-1). and . Static deﬂection of a simply supported beam. The value of the maximum deﬂection of the beam centre in equilibrium can be computed analytically [32]. We have chosen to verify the solutions for the following cases: 1. . 3.moment of area.length of the beam. which for a round cross-section with Figure 6-1: Distributed gravity load acting on a simply supported beam. 2. distributed load) the maximum deﬂection is given: where is the mass of the beam. First eigenfrequency of a free beam. For our case (both ends simply supported. 60 ã j @ } { qã |x ' & j (6-1) } . . 6. Thin ring under symmetric point force load.Youngs’ modulus. Eigenfrequencies of a free ring. 4.Chapter 6 Veriﬁcation In order to verify the theory described in this thesis some simple test cases with known analytical solutions had to be selected. .gravity constant.

Theoretical solution for the maximum radial deﬂection value exists for this load case and is given by: ~ j~ % % ` á ~ ø v já & ø v v t j ß x % % y ` ó ' n3 g3 ã x a ' 61 g|m radius is given by Ü x } Ü | (6-2) (6-3) (6-4) . two-waves in circumferential direction and the 4-th order polynomial for the axial direction. The forces are applied at the middle of the cross-section one from the top of the ring downward and other from the bottom of the ring upward (see Figure 6-2).2 A Thin Ring Two symmetric constant forces act opposite to each other in the plane of the ring. The error in the model can be measured in percents as We were interested to get error below one percent.Figure 6-2: Symmetric constant forced acting on a thin ring. To get the correct results for the complete 3D model with volume integration we had to use second order polynomial in radial direction. Note that the same test cases were veriﬁed for the beams with the rectangular cross-section where Chebyshev polynomials were used all three coordinate directions. 6.

Such simpliﬁcations becomes inaccurate for the higher eigenshapes in thicker rings and consequently leads to deviations in eigenfrequecies solution. For the veriﬁcation we used model with second order shapes in radial and axial directions and Fourier series up to 8-waves on circumference.5%. .stands for applied force in radial direction. The total number of ﬂexible states used in the model was 460. Free body eigenfrequencies in the plane of the ring are given by: (6-5) where is used to compute th eigenfrequency. Young modulus. We believe that the reason for that is the simpliﬁcation used in theoretical approach for thin rings that only allows cross-section rotation. errors in eigenfrequencies up to the 8th in the plane of the ring were also below 1%. The error in static solution became 0. We used only linear term in radial direction keeping second order in axial direction and extended Fourier series up to the 30 waves. 62 ã g à ¡ ª ` 3 3 ã ¡ f à à . The total number of elastic coordinates in the model become 1189. The relative error in the highest eigenfrequecy was 2.width of the ring cross-section in axial direction.6%) but we were able to calculate up to the 30th eigenfrequecy. The model resulted in a higher error in static solution (8.medium radius. . . Due to the large number of states required for veriﬁcation at higher eigenfrequencies we had to reduce the order of polynomials used in our model.where . We have to mention that the solution close to the theoretical one for the higher eigenfrequencies could be achieved only for very thin rings (cross section width of about 1% of the ring diameter).thickness of the cross-section.3%.

start time. The main components of the complete simulation system are shown in Figure 7-1 and described below: 63 . This overview is quite general and is applicable to most multi-body systems. Arrows denote information ﬂow. state Solver start state RHS. 7. parameters Simulation parameters. The ideas behind the object-oriented implementation of a multi-body simulation system were developed in a close collaboration with the ObjectMath project [27]. The BEAST system uses object-oriented design and programming concepts. time & state model type create for output input data Model output data Input/output initialize time. not only to the BEAST system. Note that the class and object diagrams presented in this chapter follow the standard UML (Uniﬁed Modeling Language) notations [14].Chapter 7 Simulation System Design In this chapter the software design of the simulation system is discussed. Jacobian Figure 7-1: Collaboration of components in the simulation system executable. The concepts described in this section do not change when transferring from rigid to ﬂexible modeling.1 System Overview In this section a general overview of the system is provided.

issuing the commands to calculate the output for some time steps and to terminate the simulation when the user speciﬁed end time is reached or an error occurs. Model This is an object created by the Simulation object. Note that there is a special mode of RHS calculation when the data is not returned to the solver but written on the disk via the input/output classes. The general Model class is inherited by all other model classes. The differences in the rigid and ﬂexible modeling conceptually affect the Model object only. It must be ensured that none of the objects rely on the constant number of the state variables inside every body in the model. Input/output These objects are responsible for the ﬁle I/O interface. etc). This object essentially deﬁnes the multi-body system to be simulated. As the simulation runs the simulation class monitors the progress. The explanation of the responsibilities and purpose of each class follows. The class implements necessary standard operations for data exchange with the 64 . Then the numerical solver is setup and the speciﬁc model object is created and initialized. start and end time. The internal structure of a model object will be discussed later in more detail. its responsibility is just to evaluate the time derivatives of the state variables during ODE’s RHS (right-hand side) evaluation and Jacobian matrix for the Newton iterations inside the solver. in a real implementation some details must be checked when going into ﬂexible modeling. It is an abstract class providing the interface to any model from other parts of the system. 7. However. Such an assumption could be used in a multi-body simulation tool with rigid bodies only for the optimal implementation of different service routines transferring data between the Model object and other components of the system.Simulation This component manages the dynamic simulation process.2 Rigid Body Model Classes The class diagram of the conceptually most important classes in a multi-body model is shown in Figure 7-2. reads in the model type and general simulation parameters (such as tolerance. Solver This is the numerical ODE solver component of the system which is accessed through a set of wrapper routines providing interface between the standard CVODE numerical solver and the system. During the setup phase the class initializes the input/output classes. From the mathematical point of view. The Model class. They provide the functionality for reading the simulation parameters and model input data and for writing the simulation output data. Flexible bodies have a variable number of state variables and therefore can not be handled by such service methods.

For instance. The methods of this class implement the Newton-Euler equations. The .n Contact cBConnection SPB 1. a Ring class contains several rotational segments. 65 g ¸@ w f g ¸@ w f . such as inertia characteristics and material properties.CtlPoint Model SD 1. There are functions evalsurfaces are normally deﬁned in parametric space uating the geometric point coordinates in the body coordinate space and functions that for any point in space can ﬁnd the parameters of the nearest point on the surface which is an important feature for contact forces calculations.n Connection FML Figure 7-2: Rigid Body Model Classes. This class has attributes.. A segment represent a continuous surface that can get in contact with other surfaces. numerical solver and with the input/output classes. The Body class.n Body 2 1. The Segment class. Speciﬁc body classes contain particular kinds of segments. The Body class is also an abstract class. The geometry of every body in BEAST is deﬁned by a set of segments.. The child classes of this abstract class are speciﬁc models containing speciﬁc sets of bodies and speciﬁc connections between the bodies. Segment is another important abstraction in BEAST.. describing those properties of a physical body that are not dependent on the interactions with other bodies.n Segment 2 BBConnection 1..

The CtlPoint class. Comparing to the original design presented in Section 7. The purpose of such a collection is to accumulate all the forces acting between the two bodies or between the body and the external coordinate system. force and moment loading.3. The position. since cBConnection can be seen as a simpliﬁcation of the BBConnection class (where there are no contacts and motion of one of the bodies is known) . as well as simple-parametric bearing respectively.The Connection class This class is a container class that encapsulates all the objects acting between two bodies or between a body and an abstract coordinate system deﬁned in the model space. The general characteristic of such a force element is its locality.2 the most important change is the transfer of the ControlPoint class from the Connection class to the Body class and generalization of the simple forces into the PointTie class. The single control point notion provided enough modeling power for the case of rigid body modeling. First.3 Flexible System Design 7. This class represent a single control point or coordinate system as discussed in Chapter 4. Only BBConnection class will be discussed later in the text. These classes represent simple force element: stiffness and damping.3. orientation and speed properties of the point are provided by the user. such as clearance and radial stiffness. The two child classes represent the two cases that we just deﬁned. The PointTie class generalizes the concept of a force element acting between two points. 7. The SPB is essentially a non-linear kind of spring that is parameterized in the typical bearing terms.1 Limitations of the Rigid Body Model Design The transition to the ﬂexible body model revealed some limitations of the original system design. The SD. the single control point in a connection between two bodies is not sufﬁcient as we discussed earlier in Chapter 4. FML and SPB classes. 7.2 Implementation of the Control Point Architecture Figure 7-3 shows the class diagram that illustrates the more general design suitable for the ﬂexible case. That 66 . Our design solutions to these problems are presented in the rest of this section. The BBConnection class represents a connection between two bodies and cBConnection class represents a connection to a coordinate system. Other changes concern the Body class design that should now implement the equations for the ﬂexible model case and work with the shape functions.

The nature of the points and the state of the bodies are not relevant for such a force element. The typical example of such force elements are springs and dampers (both translational and rotational)... It is essentially a nonlinear kind of spring that is parameterized in the typical bearing terms. orientation and speeds of the two points that the PointTie connects. the resulting force and moment at a given time instant depend only on the relative position. The PointTie class is therefore an abstract class with a single abstract method that evaluates the force and moment acting at the connected points for the speciﬁed relative motion of the points and time.n Body 2 1. In the context of the BEAST tool.n 2 Segment 1. 67 .. such as clearance and radial stiffness.2: SDPointTie encapsulate the force and moment calculation for general springs and dampers connecting the speciﬁed points. The generic class is responsible for the necessary coordinate transformations and numerical differentiation of the forces for the Jacobian calculations. FMPointTie generates forces and moments that are pure functions of time and do not depend on the relative motion of the points..Model 1.n CtlPoint 2 Contact 1.n Figure 7-3: Flexible Body Model Classes.n Connection 1.n PointTie 1. is. SPBPointTie (Simple Parametric Bearing tie) normally used to connect centres of two bodies that are joined by a bearing. FML and SPB classes discussed earlier for the rigid body system in Section 7... the PointTie abstraction was realized in the three child classes that correspond to the SD.

The strain operator Therefore the shape function can be designed as an abstract interface class with only two methods as described above.. The new tie is similar to SPBPointTie. but has a different set of parameters. For instance. This class actually can be used in the future to interface the shapes and integrals generated by other systems. .e.g. via the data structures deﬁned in the FEMMBS tool [28].3 Object-Oriented View of Shape Functions During a simulation the shape function conceptually only need to provide the following output: The shape matrix for any given point in the body coordinate system according to Equation 2-2. generate 68 t Þ t deﬁned in Section 2. It cannot.5. however. 7. Many numerical integration schemes. The IntFlexBody class deﬁnes the initialization step for the FlexBody class with the volume integration procedures based on weighed integration points. For the purposes of this work. The most general FlexBody class implements all the time dependent equations described in Chapter 2 and the analysis on the single ﬂexible body as discussed in Chapter 5. e.3. cylindrical or Cartesian) and use different kind of evaluations.4 Flexible Body Class Design Figure 7-4 shows the class diagram illustrating the design of the body class in the ﬂexible modeling part of the system.7. CartShpFunc is the Cartesian shape function implementing the series in Equation 3-2 and the differentiation operator according to Equation 2-29. Note that once the integrals are available all the analysis can be done on the abstract FlexBody class without accessing the child objects.The obvious beneﬁt of the generalization of the simple forces is the ability to add new kind of force elements in an easier way without concern for the rest of the system. 7. be used directly since it relies on the child classes for the initial calculation of the inertia integrals as well as stiffness and damping matrices. a new HydroStaticBRGTie was recently introduced into the system. two speciﬁc classes of the shape functions were deﬁned: CylShpFunc is the cylindrical shape function implementing the series deﬁned in Equation 3-1 and the strain operator calculations deﬁned by Equation 2-30.1. including the scheme described in Section 3.. Speciﬁc shape functions can actually work in different coordinate systems (i.3.

Even though the model is very small all the important elements of larger. More classes are needed in a real system to accommodate for the variety of different geometries. The outer ring is guided relative to the cB system by use of the stiffness and damping forces cBbER‘SD between the control points cB‘ctl and bER‘ctl.4 A Simple Model Example This section gives a simple example of a speciﬁc model consisting of two rings as depicted in Figure 7-5.Body FlexBody IntFlexBody ShapeFunc FlexRing FlexSqBeam CylShpFunc CartShpFunc Figure 7-4: Flexible Body Classes. 7. The descendants of the IntFlexBody class are speciﬁc body classes where geometry is deﬁned by a set of segments. The initialization procedure of the IntFlexBody calculates the integrals deﬁned by Equations 2-9 to 2-13 using such sets of integration points. The responsibility of these classes is to produce the set of integration points required by the IntFlexBody class. The rings have a potential contact between the outer segment of the inner ring bIR‘s1 and the inner segment of the outer ring bER‘s3. Figure 7-6 presents an object diagram for this model. a set of integration points with weights. The model is called TwoRing. more complicated models are present. 69 . The TwoRing model consists of two bodies: the outer ring bER and the inner ring bIR. Note that only two child classes are shown in the class diagram in Figure 7-4 just to provide an example of the most simple kinds of bodies.

1 RHS Evaluation The evaluation starts by transferring the state variable values from the solver to the objects realizing the Body class. Absolute position and velocity is not important for the interaction force calculations. and the newer models with ﬂexible bodies. 7.5 Solution Procedure As was mentioned in the beginning of this chapter the Model object is responsible for the evaluation of the RHS vector and Jacobian matrix during the solution of the system of ODE describing the motion in the multi-body system. are specially addressed. 7. At this point the complete state vector as seen by the CVODE is decomposed into variables representing the motion: position. However. orientation and velocity.bER bIR bER‘s3 bER‘ctl cB‘ctl bIR‘s1 cBbER‘SD contact bIRbER‘s1s3 Figure 7-5: A Simple Multi-body Model. That includes the terms caused by centrifugal forces and coordinate system transformations. The differences between the models with rigid bodies only. Having the motion state data the Body objects perform the calculations independent of the external loading.5. In the newer system design the motion of all the ControlPoint objects can also be calculated at this stage. Note that in case of rigid bodies all the forces from the interaction between the two bodies only depend on the calculated relative motion. 70 . This section discusses the mapping of the dynamic equations of motion onto the software classes and the computations ﬂow between the classes. Each Connection object calculates the relative motion between the two bodies the object connects. Th next step is done by the objects of the Connection class.

accumulate them. Finally the vector of derivatives is assembled by the Body objects and send to the numerical solver.SD: SDPointTie bIR: FlexRing bIRbER: Connection bIR‘s1: Segment bIRbER‘s1s3: Contact bER: FlexRing bER. and use the Newton-Euler equation to calculate the acceleration tensor. All the (generalized) forces acting between two bodies are accumulated by the Connection object before they are passed to the Body objects. The resulting force for a ﬂexible body always depends on the elastic state variables.5. add the contribution from the centrifugal forces calculated earlier. These forces and moments are transformed to the local coordinate system of every body. The RHS evaluation is completed. 71 . These variables are then written to the simulation output ﬁle. this is not the case for the ﬂexible body interaction.TwoRings: Model cBbER: Connection cB.ctl: CtlPoint cBbER. Therefore it can be said that the relative motion between to ﬂexible bodies can be described by the relative motion of the ﬂoating reference frames and the elastic state variables of the both bodies. In the case of output RHS evaluation some extra output variables (such as power dissipation in a contact) may be computed at different stages of the RHS evaluation algorithm. Note that transformation to the local coordinate system for a ﬂexible body includes the evaluation of the elastic part of the generalized force vector as discussed in Section 2. Body objects receive the generalized forces from the Connection object. For the calculated relative motion variables the force elements in the Connection object and corresponding Contact objects evaluate the forces and moments acting between the bodies.ctl: CtlPoint bER‘s3: Segment Figure 7-6: Object Diagram for a Simple Speciﬁc Model.

7. These objects are also calculating the mixed sub-Jacobians. q x § § l B û .2 Jacobian Evaluation The Jacobian calculation follows the procedures described in Section 2. For the partial derivatives matrices it leads to the necessity to deal with arbitrary size matrices. The analytical part of the Jacobian is also evaluated by the objects of Body class. if an object is a Connection between bodies and then it is responsible for the evaluation of the following matrices: . The use of chain differentiation rules leads to the clear separation of several stages in the evaluation where at each stage a different partial derivative used in Equations 2-79 and 2-80 are calculated. To resolve this problem the the Body class in the ﬂexible case collaborates with the Connection class to implement the algorithm for the evaluation described at the end of Section 2.8. . of 72 © §B C © £ 4§ l C ²7 ³7 7 §B C 7 £ 4§ l C ²© à³© d ©BC 7BC 7 A § § C © A §§ C £7 ²© § r C 7³ ©£ ² § l C © ³© © £ ²7 w C ³© §r§C 7 £ ² l C © ³ ©© § § C d A . and The numerical differentiation algorithm with respect to the relative motion variables is implemented in the Connection class and calculates the most computationally expensive derivatives . The rigid body model uses the same number of states for every body in the model whereas the number of ﬂexible states varies for different bodies. There are also some technical implementation details that need to be mentioned.5. The instances of the Body class are responsible for evaluation of all the partial derivatives which involve the particular body states. That is. . The following matrices are calculated in a body : and The instances of the Connection class are responsible for the partial differentiation of the forces acting between the bodies with respect to each of the two involved bodies state variables. The software developer should therefore carefully consider memory allocation and data transfer issues.8. Recall that one of the main differences of the ﬂexible body model compared to the rigid body model is the dependency of the applied force or moment to generalized force conversion from the ﬂexible body states leading to a more complicated partial derivatives evaluation.

The grinding contact needs 73 . Detailed and accurate geometric description is needed to model all the local variation in geometry of un-ground rings. In fact. a general MBS modeling framework is necessary. continuous development and research in the grinding process is needed. The purpose is quality improvement and cost reduction.1 Overview The results presented in this chapter were earlier presented in the article ’Dynamic Simulation of Grinding with Flexibility and Material Removal’ [12].2 Need for Grinding Simulation The grinding process is important in many high precision manufacture industries. Since the grinding machine is a complex dynamic system. First. For several reasons it is suitable to study the grinding process with simulations. Thus.Chapter 8 Dynamic Simulation of Grinding 8. most machines cannot be dedicated to research activities for sufﬁciently long periods of time. For SKF as a manufacturer of high precision rolling bearings it is about a third of the total manufacturing cost. The dynamic behaviour of the grinding machine is primarily determined by the contacts between the workpiece and parts of the machine (various supports and the grinding wheel). Simulation of the grinding process requires simultaneous modeling of several phenomena including: Multi-body system dynamics. as well as the geometry changes during grinding. Second. Tribological contacts with material removal. this was the ﬁrst industrial application of the developed technique. 8. Hence detailed contact modeling was necessary. in order to perform the simulation. The approach presented in this thesis was used to model the workpiece ﬂexibility. it is very hard or even impossible to measure some characteristics of such a dynamic process as grinding.

There are at least two shoes that support the ring and the microcentric forces are holding the ring in position. and driven by friction forces. so efﬁcient computation is essential. special treatment since it results in material removal from the workpiece. 8. some parts of the machine can be reduced to a fever number of important bodies for which the movements also can be measured and analysed. thus changing the original geometry of the body.g. with a magnetic force. Microcentric arrangement means that the workpiece (bearing ring) is clamped to the drive plane.Figure 8-1: A model of the grinding machine including most of the parts of the machine. The models are used in complete grinding cycle simulations. However. The examples presented in this chapter are concerned with microcentric grinding. Hence in such cases the structural deformation must be included. e. The drive plane centre has a small offset distance (microcentric) relative to the ring. Initial stress distribution in the surface layers in combination with uneven material removal will cause distortion. Distortion. Microcentric ring grinding is one of the most common grinding processes in SKF and so the investigations were started with this process. 74 Elastic deformation. The stress distribution is created by material phase changes in the hardening process. . One example of a complete model of a grinding machine is shown in Figure 8-1..3 Grinding Machine Model A grinding machine is a complex and large system consisting of many components. Large forces are applied to the workpiece during grinding and the structural elastic deformation of thin rings is in the same order of magnitude as instantaneous material removal.

The contact is in a different regime depending on the gap between the rings: The ﬂuid ﬁlls the gap. a pure dynamic analysis is satisfactory. Further. If only the effect of geometrical ﬁltering is of interest it is enough to use pure kinematics. compared to what is common in other grinding simulations [2]. Detailed definition of the geometry of this arrangement is crucial for the correct modeling of machine dynamics. This machine has few major eigenfrequencies and so a limited number of bodies can accurately describe the dynamic behaviour. With smaller gap the ﬂuid pressure causes some elastic deformation and the contact is going into the beginning of the elastohydrodynamic (EHL) regime. It is the hydrodynamic regime and a pressure is built up which gives a rolling resistance moment and normal force. is the use of fully three-dimensional structures. Small material damping due to rolling resistance also shows up. Shoe supports are also responsible for some ﬁltering effects that minimise the effect of certain imperfections of a ring. Even a small deviation can result in malfunction of the arrangement with microcentric forces pushing the simulated ring out of the machine. Both of them have constant but different rotational speeds.4 Modeling Tribological Contacts with Material Removal In this section we will outline the most important phenomena that can be observed in a grinding contact. Flexibility and stress analysis are also very important if effects such as stress release of internal stresses during grinding are of interest. they are small due to the low viscosity and large gap. But if the purpose is to simulate the total grinding cycle and the effects of machine dynamics on the ﬁnal output shape. It is important to mention that different complexity levels of the model should be used depending on the purpose of a simulation case. If instead the machine stability and chatter are the only interesting phenomena. A measurement of the dynamic response of the grinding machine can identify these dominating eigenfrequencies. There is a rich supply of cutting ﬂuid (mixture of oil and water) at the contact inlet. The load will be shared between the asperities and 75 . However. without impact) we move the two rings toward each other. Slowly (i. The shoe support arrangement is an essential part of the machine. However. 8.In our example an external ring grinding machine has been modeled. Consider a rotating solid grinding wheel and a rotating ring. One signiﬁcant difference of our models. a complete model such as the one suggested here is necessary..e. in the case with a grinding wheel that have quite a rough surface we will quite early get asperity contacts. it is of importance to include the ﬂexibility of the bodies in order to get accurate results.

76 The ”gap” becomes negative.. However. Our approach is on the macroscopic level. surface geometry changes. the material removal rate is normally increasing faster than the other parameters. The most typical problem encountered in the ﬁnal product is waviness of the ground surface. At a certain load the grains of the grinding wheel have sufﬁcient pressure to start to cut. All the physical phenomena mentioned above are modeled and included in the tribological used contact model. squeeze ﬁlm build up. Another trend is to have quite simple macroscopic models. With our simulation tool the shape of the workpiece can be animated for the entire grinding cycle. 8. there is an intersection. consequently. and it is possible to see how the shape develops during the process. . 25]. In the literature one can ﬁnd two directions regarding grinding contact modeling. Each of these phenomena is a non-linear function of load. They will result in additional pressure/forces. However. we include some important non-linear behaviour in the model to have good correlation to our experimental data. due to the low viscosity the ﬁlm is very thin and nearly all load is carried by the asperities. it is the shape of the ground piece that is one of the most important outputs. elastic deformation. i..g.5 Grinding simulation results The output from a grinding simulation can be any type of forces and movements. Grinding simulation has now been used for a wide range of investigations and some example studies are presented below. e. The elastic deformation is signiﬁcant and we are well into the EHL regime. However. and material removal due to cutting. material damping. We have large friction forces. in many cases linear. The asperity contacts also result in a fast growth of friction forces.e.the ﬂuid ﬁlm.g. However. e. One direction takes a very detailed look at the cutting of individual grains [31]. etc. this is not suitable for dynamic simulations.. viscous damping. With even more negative gap (larger load). Three different mechanisms are active: some viscous rolling resistance with very thin ﬁlm building up. We also include the EHL part. for the material removal [17. Besides. In a more general dynamic situation the velocities normal to the contact surface must also be considered. cavitation. local deformation and ﬁlm thickness. The cutting will result in a certain amount of the material removed and. the ﬁlm build up and material removal are also non-linear functions of speed.

Such a machine will produce much better. were the form of the ring and a FFT analysis of the waviness of the ring is shown. rings. can be seen. as well as the shoe support system. would make the machine stable. Chatter Chatter is a regenerative wave creation on the rings. A grinding machine that is not stable enough will produce non-circular. These effects have been investigated with this simulation tool. Another example is shown in Figure 8-3. In Figure 8-2 a snap-shot of a ground ring is shown. The unbalance will create an excitation frequency into the system depend77 . It interacts with the dynamics and eigenfrequencies of the machine. As the grinding continues these waves becomes bigger and bigger in a regenerative way. wavy. In the beginning of the grinding cycle small waves starts to grow on the ring. even if the grinding machines have automatic balancing units.Figure 8-2: Microcentric shoe support system. The material removal height is shown by color. If the unbalance is too large the waves will become big and the ring will not be approved. If there is a small unbalance this will immediately have an effect on the ring. rounder rings. Unbalance It is very hard to get perfect balance of the grinding wheel. It is not only the unbalance itself that is dangerous to the grinding. This wave number could be correlated to the ﬁrst (lowest) eigenfrequency of the machine. The simulations also showed that just an increase in the stiffness or the damping of the machine to a certain level. The colours on the ring surface show the waviness of the ring at this time instance. This is an example of how this simulation tool can be used to investigate and improve grinding machines with instability problems. It is noticed that the characteristic wave number is 23. The ring and part of the grinding wheel.

Figure 8-3: The shape of the ring and FFT analysis of ring waviness. 78 .

eigenfrequencies and geometrical ﬁltering. 79 .ing on the speed of the grinding wheel. Thus. For such a system there will be a ﬁltering effect. it is very important to avoid an unbalance that is interacting with other wave creating effects. These wave numbers are depending on the position of the shoes in the shoe support system and of the contacting angles of these shoes. but it also means that another waviness may even be magniﬁed. Those were unbalance. Wave creation on the ring From these investigations it was seen that the particular machine had three major effects that affected the waviness on the ground rings. which means that some waviness on the ground piece is suppressed. It was also seen that it is very critical if any of these effects are allowed to interact with each other so that they together create the same wave numbers. Feed rates Figure 8-4: Material removal height for the models with rigid ring in one case and ﬂexible ring in the another one. and if this frequency lies near some of the eigenfrequencies of the machine. Thus it is very important to be able to predict and eliminate such interaction when the machine and its operating conditions are designed. large vibration problems will arise. Geometrical Filtering The grinding machine that is exempliﬁed here uses a microcentric support system.

For rings that are not very stiff it is important that the ring ﬂexibility can be included in the simulations. When the grinding wheel is fed toward the ring during grinding a normal force arise between the workpiece and the grinding wheel. In Figure 8-5 the deformation of a ground ring can be seen (magniﬁed) as well as the effective stresses that are shown as a colour diagram on the ring surface. An example is a set of simulations to ﬁnd optimum feed rates for the grinding cycles. 80 . However when using ﬂexible ring model an extra effect comes into play: the workpiece gets deformed which means the same feed rate of the grinding wheel results in lower forces in the grinding contact and it will not accelerate. Flexibility is also of great importance when the effects of stress release during grinding are investigated.Figure 8-5: Deformation and stresses of a ring that is being ground. The deformation is magniﬁed. This example clearly shows the importance of ﬂexibility in grinding modeling in some cases. The ring rotation speed starts to grow when the drive plane loses its grip due to the driving force from the grinding contact. At a certain force level the friction in the grinding contact overcomes the friction on the drive plane for the rigid ring. In Figure 8-4 the material removal is shown for a case with drastic difference between a rigid and a ﬂexible ring. As the ring rotates faster the relative speed in the grinding contact becomes smaller and the material removal rate goes down.

8. The tool enables detailed studies of a complete grinding cycle. as well as models for material removal have also been implemented. Detailed models for the contacts between the workpiece and other parts of the machine. It provides an efﬁcient way to investigate possible optimisation of both the machine design and grinding process parameters. 81 . Completely three-dimensional modeling is used. The tool is now actively used at SKF for parametric studies of different grinding schemes.6 Summary A simulation tool capable of dynamic modeling of the grinding process has been developed and the author contributed to the development with the implementation of the structural elasticity model. The process model used in the tool covers most of the important factors of the grinding process. All the essential components of a grinding machine are modeled as a multibody system.

Finally. The procedures necessary to perform these types of analysis inside a dynamic simulation tool were presented. Some approaches to simplify these equations by choosing the coordinate system for the solution of the equations were discussed. The system design that provides the necessary features to specify the external boundary and loading conditions was discussed. Special attention was given to the analysis of the generalized force resulting from the residual stress release during grinding. generalize viscosity forces (with special focus on Rayleigh damping). The penalty approach for enforcing the boundary conditions was chosen as the most suitable method for the discussed case. The thesis 82 . In particular generalized elastic forces. Additional analysis types (eigenmode and quasistatic) for ﬂexible bodies that are normally found only in ﬁnite-element packages were discussed. The generalized force vector was deﬁned and procedures to calculate the generalized forces for different kind of loading were presented. The developed model was implemented in a real simulation system. The necessary boundary conditions for the proposed mode shapes were discussed and the methods to enforce these boundary conditions were compared.Chapter 9 Conclusions Dynamic equations of motion of ﬂexible bodies when using the ﬂoating frame of reference formulation were presented in this thesis. Some examples were used to demonstrate how complicated loading conditions can be easily speciﬁed using the proposed design. The advantage of using the well known mathematical functions in terms of calculation efﬁciency and ease of use were demonstrated. some of the single body analysis procedures were used to verify the accuracy of simulations when the proposed set of deformation modes was used. generalize external force and body load were presented. The veriﬁcation also provided an insight on how to select the number of functions in the function series that are used as the assumed mode shapes. The proposed mode shapes use well known series of mathematical function as the basis for the modes set. A set of assumed mode shapes for modeling deﬂection of bodies with relatively simple geometrical shapes was introduced.

The system has been successfully used in industrial simulation of the grinding process [12]. 83 .describes the system architecture and class design. The issues that are speciﬁc for the ﬂexible multibody systems as compared to the system with only rigid bodies are specially addressed.

More investigations are needed to improve the performance of these calculations.Chapter 10 Future Work One of the important features of the BEAST system is the ability to run on parallel computers. However. The possible candidates include spline functions and wavelets. the use of ﬂexible bodies in real models has uncovered some limitations of such a straight forward transition. The operations on a single ﬂexible body with large number of states requires manipulation of large matrices and vectors. The approach with the general shape functions that include Fourier series and Chebyshev polynomials proved to be efﬁcient for the grinding simulations. The main idea of the parallelization is to deﬁne the contact forces evaluations as computational tasks and use master-slave approach to compute those tasks at each RHS or Jacobian evaluation. The contacts with ﬂexible bodies are more computationally expensive. Hence to improve the scalability of the system to the larger number of processors such ﬂexible contact calculation tasks need to be decomposed into several smaller sub-contacts. That means the amount of time spent in linear algebra computations grows very fast. The algorithms used by the system were published earlier [13]. In many systems only few contacts involve ﬂexible bodies while the rest of the system is composed of the rigid bodies with cheaper ’rigid’ contacts. Since the transfer to the ﬂexible modeling was done with relatively small changes to the overall system design the same algorithms could be used to perform parallel simulations involving ﬂexible bodies. Therefore the time to evaluate a single ’ﬂexible’ contact deﬁnes the lower time bound for a single RHS evaluation effectively limiting the possible speedup on a parallel computer. 84 . The question that remains open is the search for other kinds of general mathematical functions that might be more suitable for different geometries.

i) generates a Chebyshev polynomial of order generates sine and cosine pair for the -th wave. Polynomial(x. . . The components of this vector.e. the deﬂection vector of the material point. . Examples of pipe deformation shapes that can be represented by the modes deﬁned in Equation 1-1 are shown in Figure 1-1.Appendix A Shapes Generated by General Functions This appendix is a supplement to Chapter 3. . . i. Note that for a realistic free 85 g @ Ì £º g h@ Ì f £º g h@ Ì f £º f (1-1) and d § ª Ì § Â » Â 3 g @ Ì ~ º R ¨n3 f Â Â » Â 3 g @ Ì ~ f º xn3 Â Â » Â 3 g y@ Ì ~ h £º ¨n3 f Â à Y g Ì k @ Ì z@ Ì ~ f| ºg @ Ì £º ºg @ Ì f º ºg @ Ì f £º f ½ ½ ½½ ½ ½ Y áY Y Y áY Y á Y á áY á Â » Â e e W¡ Â » Â e dW¡ Â » Â e W¡ w Y % õ áY á e d¡7 ¡7 X¡7 ½ Y w @ áY w e ©¡ g Ì w d©¡ g Ì § f ½ Y w ©¡ g Ì § f á Y w §fY g @~ £º ¨ f Â § Ì § ¤¤¤¤æ ä¤¤¤¤ å . represent deﬂections in radial. . The resulting position of the material point is obviously: . It provides some basic examples of the general shape functions deformation modes for the case of a ring structure. and . That is in cylindrical coordinates it has components: . and are the nine parameters of the shape functions that specify the length of each series. The following parameters should be speciﬁed to the series in order to enable these shapes: (a) Ring expansion/compression is a deﬂection in radial direction as a function of radius. tangential and axial directions. . The most general representation of the deformation shape functions used in BEAST for rotation symmetry bodies can be written as: where vector represents undeformed position of a material point. Setting can give such shapes.

(d) Ovality mode of a ring. (e) Pipe torsion. (a) Ring expansion/compression. (c) Taper like deformation. (b) Rotation of ring cross-section. (f) Conical deformation. 86 .a) b) c) d) e) f) Figure 1-1: Examples of deformation shapes.

(e) Torsion of a pipe is enabled when both and . (f) Conical deformation shown in the ﬁgure is a combination of deformations enabled by non-zero parameters and To give a real usage example let us assume that we need to simulate a relatively stiff ring. and are greater then zero. . (b) Rotation of ring cross-section is a result of tangential displacement with respect to the radius and angle. (c) Taper like deformation can occur when axial deﬂection is a function of radius. (d) Ovality mode of a ring.symmetry with radial parameters to get stress- free central line .allow axial expansion The parameters specify a model which is stiffer than the real body but can simulate the ovality mode and will run fast due to the low number of shapes used. 87 Y ½½ Y ½ õ Yá ½ Y% á õ á Y á ½½ % õ Y ½ Y ½ % õ áY Y ½ áY á Y % ½ ` Y % Y @ % áY ½½ ½ ½ Y @y Y @ ` á Y % ½Y á y Yá áY á 7½ ½ ½ y Y % õ áY y % õ áY ` . The minimal reasonable combination of the input values will be allow radial expansion/compression two waves correspond to the ovality mode for a narrow ring taper shape is not important . . In practice it is often worthwhile to try different combinations of shape functions parameters and check the results of single body analysis described in Chapter 5 to select the minimal number of modes satisfying the particular simulation requirements.ring radial expansion mode is normally coupled with changes in axial direction so that ring volume does not change under the deformation. In general to simulate -waves deﬂection must be at least equal to . That is why parameter should also be set greater then zero for this case. To get this mode one should set . Set to enable it.narrow ring has practically no deformation of this kind . We only expect ovality mode from it and we don’t expect much deformations through the cross-section. Setting and enable such deformation.

ed d H w x 5 0 C ¤EI¤ctt¤¢¤P 2¤uwtuwXxvt y¤¢¤P 3 x w x 5 0 C Th F w 2 5 C P ¡ ¦ 4igf vvf ¤0¢¤`Y¢C ¤ HIu¡¢C ¤tY¤¤¤¤B F ¦ § ¦ ¡ ¥ s r HI¤qY¤¤¤¤YXS¢¤¤¤`g¤Y4ig¤¤edV¢b ¢6 ¤S¢¤ F F C & § ¨ 2 R Q P p a T h F f f C a ¨ Q HIc¢b a¢6 ¤¢¤ `Y¤¤¤¤B F V ¨ S Q § ¦ ¡ ¥ H ¡ § X¢) ¤VWTU¢¤¤¤ S R Q P H F 8 C ¥ 0 B 5 ( 2 0 5 ¡ ( ( IGE9D¤¢¤) ¤¤¤¤¤¡¢(¤¤4A@98¢7 ¢6 ¤¤¢¤¤4 3 2 0 & ¦ ¥ ) ¦ ' § & ¦ " ¦ ¡ ¥ § ¦ ¡ ¦ © £ ¦ ¨ ¡ § ¦ ¥ £ ¡ ¢1"#¤¤¤¤¤¢¤( ¤¤¢$ ¤¤¤¤¢$%#¤§¢! ¤¤¢ ¤¤¡¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¢ .

eà á ½ ¼ » ¤¤Yz¢Úd¤¤d¿¢¾ p¢ ¤¢¤º v u k Ô k Ôo v © Ó Ó n Ó n u © u ¢o t j n u o q w ¢}¢#ÙXy DAzIu¦v¤¤yvzIc¢£g¤¤n Ã Ã Ä ÃÃÃÃ Ã Ã Ä ÃÃ z © z u ¢ t j n z u ¢o q w ¤9ßÞu£oY¤¤Ù}Þu£Y¤¤Ùn X¤l¢ ¤tg¤¤r¢ ·{ Ã Ã Ä ÐÃÃÃÃ Ï ÍÎÌØÉ ÇÀ ÃÃ Î¢À z ¿o n l Ý Á Ë Ê À È Æ Å Â Ä ÃÃ Á Ï ÑË ÒÍ ÌÊ Ü¤¤ Â k ½ ¼ » v u ² Ô ² Ôo IÛ¤vz¢Úd¤d¿¢¾ p¢ ¤¢¤º Ã Ã Ä ÃÃÃÃ Ã Ã Ä ÃÃ v © 9Ó¢#yXÙ#9AIc£Y¤¤Ùvu£oY¤¤wÙn { Ã Ã Ä ÐÃÃÃÃ Ï ÎÌØÀ Æ À ÃÃ ¤Á@}×¤ Ó n Ó n © z u ¢o t j n z u ¢ q Í Á Ë Ê È Å Â Ä ÃÃ À ² Ï Ë ½ ¼ » vIz¢ÖÍ ¤Ê¤¢Õd¿¢Â ¾ p¢ ¤¢¤º u Ô Ò Ñ m Ô o v s { gyÖ ½ ¼ » ¢Ô¤l¢¤ ¤¢sÕd¿¢¾ p¢ ¤¢¤º v u s z k t Ôo Ã Ã Ä ÃÃÃÃ Ã Ã Ä ÃÃ v © Ó ¢ Ó n Ó ¢}#yn Ã Ã Ä ÐÃÃÃÃ Ï ÍÎÁ Ì|É ÇÀ ÃÃ ¤@} Ws Ë Ê À È Æ Å Â Ä ÃÃ Á À Ï ÑË ½ ¼ » vI¤ª¸#toY¤td¿¢¾ Â p¢ ¤¢¤º u u © z ¢ ÒÍ z ¨n Ê w o µµµ ° ° ´µµµ ² ¯¯¯ vID¹¢¸#nt·x w ¶ ° u¦Y¤¤G± u£oY¤¤j 4id«ª}¨¥z}y toY¤w u © z z o u ¢o q w u ¢ t ¬®¯¯¯ u y © z y ¢ n ³ ° u£Y¤¤j u£oY¤¤w u ¢o t u ¢ q z u ¢ q w q § u ¢o t j u¦v¤¤¥ ¤u£oY¤¤¡ ¤¤¢¤ ¤¤m ¤¤¤¤¤¤¤¤¤¤¤¤¤¤l¢x m¢ ¤¤¤¤¤¢x ¤¤¤¤¤j l r q m l r n m j q q q t l n n t ¤wvgzY¤¤¤¤¢ vX¤¤¤¢ Yz¤¢ }¤¤}vX¤¢v¤¤ z z s z x z k n m z k n m w n m w z k z z z g¤¢ ¤¢ 4iYy Y}|¤ Yz¤¤y vz}|dv¤x toY¤¢s ¤¤p l k t u z y z y y z ~ y { z y w k t r q vIc¤¢s ¤¤tY¤¤¤¤j u k t r q po n m l k i h f f g .

qo p ø n ø n ø k ø n ø Þ896X6# #6Xn6ø#42 n ø n ø n ø ø ¤89qq#n#klH m jmx2 j ø ø j ø 8 n n ø ø j ø k ¤9qøq#nmXlH jm x2 ø j ø n ø ø n ø j ø j ø ¤89qnq#mm#klH x2 j 2i×ö h æ D % D G ö å ä f g 1f h£# å " ä $ " £ä " ££££ä # ¤ " ¢g xe ý û û ú ø ù ø å÷ R ø © å I¤A}#t·¥â þ @d¤G ä å R å ä © §¦ ¢ ÿ " ¤¤¡¥Gâ þ ¤¤£G ¢ ÷ ¨¡¥¡¤¥£¡¤þ ý ) 7& û ú ø ù ø å R Y('D·}#t÷·¥â 0(1DA}#t·R¥â ) 5& û ú ø ù ø å÷ å ø 0(1DA}#t·R¥â ) H& û ú ø ù ø å÷ 0('Ißø}#t¥â d ) 7& û ú ù ø å÷ R å å ø ) H& û ú ø ù ø å÷ R Y('D·XtA¥â ) 5& û ú ø ù ø å÷ R Y('D·XtA¥â d S0(1DA}#t·R¥â ) H& û ú ø ù ø å÷ ø0(1DA}#t·R¥â ) 7& û ú ø ù ø å÷ å å ø ) 5& ú ø ø å÷ R 0('ûAù}#t·¥â ) H û ú ø ù ø ÷ R 0(&DßÛy#å A¥â B ø0(1DA}#t·R¥â ) 5& û ú ø ù ø å÷ æ ú ø ø ÷ 4ûY·ù}}¥åt·R¥â þ us vtr ý û F £F $ ¢y£F $ x÷ ¡¥¡¥£¡¤þ 0X0(X££¤ä¢ã â qÜ£F $ ø © F © ¨ §¦ ¤¢ ÿ ý ) & ) & å p æ F w ¤ ¤ © §¦ ¢ ÿ ý ¤ ¤ ø Iû ££$ ¢© ££$ ÷ ¨¡¥¡¤¥£¡¤þ ¤ ¤ ý û 7 h Ii÷ £# å " ¢g¥I " # " £ £e$ æ ££$ ä D f ¦ ¤6££¤¢ã ` ÷ $ ¤££ö ££# " £ # ¤v¢C££¤¢ã ` ÷ ¨¡¥¡¤¥£¡¤þ ý û û H ø å ä å G F # E D ö â ø © å ä © §¦ ¢ ÿ ýd89#6X43Þ96#5#4@¢}#Ùå0'DA#å v¤â A 7 ø 5 ø 2 ø 8 7 ø ø 2 ø ú ü ù ü ü å ) & û ú ø ù ø ÷ ö ê ê ë êêêê ê ê ë êê ) & ú ø û ù÷ % G a å û ù÷ Y(18c·Iu£g££bYøIutvD # yå 2 õcë óÒêêêêê ò Ìð ÉÎí ì ç ê é ë êê è %£¡ã ä ¨ Ü££¤¢ã ` â ç æ å ä èñ çî ñ ýð û 8 7 ø é 5 ø 2 ø 8 7 ø óÒò ôÌ¤û 9#6X43Þ96X56ø#42 $ ø û ) & å ä ÷ % G R% D ä I¤û 0(&X)Y(W££¤¢ã â d££¡¥££¤å ¨ ¤æ Y(X0W££¤¢ã tdI £# ¡R ¥# â æ ) & ) & å ä â÷ E g£¡ã ä ¨ ¢V£ST¤QÛQg££¤ä¢ã tg££¡¥££¤å ¨ ¤æG ÷ ø © U û û ÷ û ÷ å â÷ % G R % D ä æ û $ I â© §¦ ¢ ÿ ÷ ÷ å ä ûÛQY££¤¢ã â×£¤¡¥£å " £$ £÷ ¨¡¥¡¤¥£¡¤þ ö Pö I ¤H6££¤ä¢ã t÷ $ ¤££ö #££# D ö â © å ä â © § ¦ ¢ ÿ ý û û ø å â å G F E " £ # ¤vø¢C££¤¢ã ÷ ¨¡¥¡¤¥£¡¤þ ýB896ø#5#4@97#56X423@üù#yåXüÙu0(&1DA}#t÷Y¤â A 7 ø 2 ø 8 ø ø ø ú ü å û ) û ú ø ù ø å ö ê ë êêêê ê ë êê ø0('Aù}#åtY¤âvX£ ¥£¨ gå¤ ¡¥ õc Òò ñÌØÉîÎíeç è¢%£¡ã ä ¨ Ü££¤¢ã â ) & û ú ø ø ÷ ö ø % $# ÷ ó è ð ç ì é ç æ å ä " ! ê ë êêêê ê ë êê óÒÎ ò é ý ô Û IñÌð å¤ä¢ã âYø¢©¤¢ã ÷ ¨¡¥¡¤¥£¡¤þ û å ä â© §¦ ¢ ÿ ê ê ë êêêê ê ê ë êê ý9ùXÙXÙåDAøù#t÷Yö¤â ú ü ü å ü û ú ø å uë Ðêêêêê ê ó ÌØïîÎeIç ê ê @æÜ¤¢ã â õ ò è ñ ð ç í ì é ë êê è ç å ä ó ôñ ÒòÎ¨ð é .

à ß ³ ¶ ¼ Ü ± x £Þ£¡¥§ÝÎ º Û£ z Õ { ² ± ³ z t ³ Ù Ú Ù Ù Ì ± 0w ¡Ã 6t Ã t ¡Ã t £Ö t ££££Ö t £££¡¼ º z z z º z Õ { ² ± Ø Á × z ¸Õ t ±££Õ££z££¡¼ ºt £££¡¼ 6t ±££££££ t ±£ t £¡ t £Á r Ä£{¡¥£¡Á w £|£y t ¥{ » £¡¼ qt ¥{ » £¡¥ t ¥£¡¥ t ¥{ £qt º º ¼ ¼ » º ¹ ¹ r ³ ³ Ö} Ô Õ z z ¹ Ô t ¥£¡´ t £££´ @t ¥£¶¡Ã t ¥£¶ t tQm££¡ª¥£££££¡ Ô Õ ª z z ¹ } z x Â££¡¥£££££Ô¡£££{ ¿ Ó x ÒÒÒ ÒÒÒ ÐÑ s Î} £Ëi£} Ã É } £ÍÌQ£} Ã Î} £Q£} Ã s Ê} £Ëi£} Ã É £ËÌÏ£} Ã É ÈÈÈ } Ê} £ÍQ£} Ã ÇÈÈÈÆÄ£m££££¡£À¥z » Å Ã} ¶ ² ± Á s ¶ ² ± Á À } z x Â££££¡£¥z » Q£££{ ¿ ¾ x ½ | º ¼ } ¸± z x w £y t { » £¡¼ 6t { » £¡¥ t £¡¼¥ t { £º§6t £¡´ t ££´ r m£¡¥£¡ » t º ¹ ¹ | ² ± z ³ ³ ² ± °} ¯® ¬ « ª x ·£y t °C ¶££££¡´µ££££££££tm¡¥¡¥£¡£ ¨ ¨ |{ ¢ | x ¡ ¨ b9©§ ~9t i}~¥£z¡y ¤ £ dt¡1 £qy£y t ¥ | } ¥| x z x t9qt Q}~¥¡¡ t9qt £¦Q¢ ~¥ ¡}££¡ |¥££££ ~t t ¥Q~|¥£¡y } { z xYw w96u6XX6u6X4r v t s t s t s t s t s t s t v t s t s t s t s Yw96u6XX6u6X4r .

fd e H H % 3 ) S G0!'1) B C£££ù¡¥Wc) B 10S!1£££ù¡¥ë ' ò ö ú ý ë 5 ' ) ' ò ö ú ý î 3 ) T' ) S' ò ö ú ýë 5 ) S' ) ' ò ö ú ý 'G0!12!C£££ù¡¥Wc0b10T!1£££ù¡¥ë % 'G) B 12!C£££ù¡¥Wc0b1) B 1£££ù¡¥ë % î 3 ' ) T' ò ö ú ýë 5 ) T' ' ò ö ú ý û H û û ¦ A ¨ õ ô ú ù @££¡ý¥ü ðÂì ¢ ££££¡¥6¡Cí ì ¢ ó££££¡¥ü¥xmè¡¥¡å¥£¡£á ï ó ô ô ú ù ýü î ê ô ô ú ù ý êé çæ ä ã â ¦ A û aH Y ð Q VUXUX4RYVTuXö(Rî Q ¨ T î S î ò ¦ î ¨ î S î ¦ Y `H H 34û ) ö ' ) ò ' ò ö ú ù ý 0(12(C£££¡¥ë û H H Y A û 2(10C£££¡ý¥ë û ) ö ' ) ò ' ò ö ú ù 340'10(Fò£££¡¥ü A 0(12(1£££¡¥ü ) ò ) ö ' ö ú ù ý ) ö ' ) ò ' ò ö ú ù ý 5 H % " ED $ "EI $ % % %% û î 9û YH `©@@ @@ 79 0(F0(1£££¡¥§63G0(12(Cò£££¡ý¥ë ) ö ' ) ö ' ò ö ú ù ý ë 5 ) ñ ' ) ñ ' ö ú ù X@ H A û @@@ 900('12(C£££¡¥W642(10C£££ù¡¥ë 7 ) ö ) ö ' ò ö ú ù ý ë 5 3 ) ñ ' ) ñ ' ò ö ú ý A û YH `X@@ H û "#D $ Y @@ 900(1(C£££¡¥ü§64(10C£££ù¡¥ü 7 ) ö ' ) ö ' ò ö ú ù ý 5 3 ) ñ ' ) ñ ' ò ö ú ý B î ò õ ö X£õ Ï¤æ ¤U£¡ý¥£Äõ@ì ¢ ££££¡ý¥ü ô ò ù è ó ô ô ú ù ð ô ù ü ê í ô ú ý êé ç ä ã â ïÍ£ô£ú£¡ý¥î¡Cô£££ù¡¥ü¥xmè¡¥æ¡å¥£¡£á ðïd¨VTîXSUîXò4¦W0¨VTUîXSuö4¦R0Ëò6Xi6ÿ¡¥£ô££¡¥üQ£ò¡¥£èõ@£ô££¡ý¥ü î î î ï î öé þ ô ú ù ý é ô ý ù ô ú ù ð Q Q H H 3 Pû 0(12(C£££¡¥ë û ) ö ' ) ò ' ò ö ú ù ý H A û 34)2(101£££¡¥ü A 0(F0(F£££ù¡¥ü A 0ö(1(C££ü 5 ò' ) ö' ò ö ú ù ý ) ö ' ) ò ' ò ö ú ý ) ' ) ò ' ò ö % % H H @@ @@ 79 (12(C£££¡¥§342(10C£ö££¡¥ë EI ) ö ' ) ö ' ò ö ú ù ý ë 5 ) ñ ' ) ñ ' ò ú ù ý û A " % $ uû î H ©@@ û @@ 980(F0('1£££¡¥§56G0(12(Fò£££¡ý¥ü 7 ) ö ' ) ö ò ö ú ù ý ü 3 ) ñ ' ) ñ ' ö ú ù % @@ "ED $ B CA " #! % &$ @@ 982(10öx(££§642(10ñx(££ü 7 ) ö ' ) ' ò ö ü 5 3 ) ñ ' ) ' ò ö Xò Ï¤æ î ö 0¨©§¥ô£¡¡ó ¢ dm¡qîq¡i6ÿ¡¥£££¡ý¥ü î ¦ ¤ ò £ ø õ í ï ò öé þ ô ô ú ù ð ï ì ë û ñ ú ì ø ÷ô ö ë õ ì ô ó ò ñ ì 0Ë£Qém££££ù¡£¥££@££££££ë ðË£¡Cm£ë¥xmè¡¥¡å¥£¡£á ï ì ë î ê í ì êé çæ ä ã â .

f¬ « « ª u{ m w x x p £ n o o i r d d #©¨4Gw yw FGw §¦ x d d ¦ ¦ l x d p i h g p k m n i p d p h p k m p n i p 8§yG|8PG¡¥G¡pUx8eiG¡¥G¡ ¥pG¡kG¡Gx8PG¡ViG¡xy¥G¡kG¡G¤ y n m m ¤ d p i h g x p h g o y r t o p y n m m x d p i h x d o y r t o p k m p n i p 8PG¡U8P¥¥G¡UxUaG¡kG¡G¤ l p y n m m p n l x r k } y n m m ¤£ n p x o G¡Gf§G¡G d x eGw GGw G¡G¡G I ED u w{ s u t s r q p i h u s u t s y r t G`|u¡10!FG¡¢¡10!1o t EI #D w u#u§1!CG¡(cs10!s1G¡g f §10b(o w { s u t s r q p i h u u t r q p i h u s u t s t Gw d x w Gw G¡G¡G i h { o k y p h g { p i h 8 uGwa{pG¡g f U w |iG¡Cw G¡g f ¥ wWw d p x p eGw GGw G¡G¡G u w{ p i h { y r t t o t p G`|G¡#|o #{|o Gw y r t { y r t t o o o u t p o q y r t d ¥iG¡UWo t i r w i t o i y i x n t o i x i r i } { o n t k g {|okGpG¡rG¡~|Gm p t h o o G¡rizoG¡jGsG¡GvGsGl y x m l k i n h w i h u t p o i l p p i G¡orkqkG¡jG¡ f h p o n n m l k i h g d p i h g p i h eG¡UxGG¡g¡G¡G¡G 0 0!12!CG¡Cc0b10!FiG¡g u s u t s r q p i h g w u t s u s r q p h xy0!12!CG¡Cc0b10v!FiG¡g u v s u s r q p i h g w u s u s r q p h x u t s u v s r q p i h g w u v s u s r q p h y0!12!CG¡Cc0b10t!FiG¡g .

¤¢ £ º 0¹ y¾ Î 0æÁGFFD§¥¥G¡Gê º ¹ Â É Ô ¾ Â ¶ ¾ À ¶ ¾ ¼ ¶ » » ³ µ ° ð ð ± ¯ ¶³ µ ² Ð ð ° ÿ &GÇ sþ1#äP¹ æÅV61¥U¥G¡¥úÔ ý ð ² ± ¸ Û ð Ü ± ¯ ° ý ü º ¶ ¹ Á ¾ Ä ¾ ²³ ¸ Æ³ ° Ü · Ç Ü µ ± ² í Ì É ÿ ý ð µ ° · ø ý ü º û Ç þGø¡GÈË1¶2ÅÌ Î º Â ¶ ¾ À ¶ ¾ ¼ 0æÁGF¿1¶!»Wú¶ Ì º8ÒùÊ©s·GãÏ°G¡° ç bíG¡ëG¡´ ç ¹ Ç ¾ î Ç É · µ · Û ð Û Ü Ç ø ² · É î ³ ì ê é è ºF#PæsVÄÈ1¥³U÷æÅ¾8P U®õ¿8E³UÆõDWôÇ É ¼ä ¹ Á ¾ ¾ ² ¸ Æ ö Â Á ¹ Ä³ Ü · ² ¾ ¹ Ä µ Û ² » Ì º ¹ 8¹æÁsVóF²¥ÆUGòG¡G¡jïbíG¡ëG¡´ ç ¾ Ä ¾ ³ ¸ ¾ î Ê ° Ç ± ñ Ü µ ± Û Ü ° ð ² ± ® î ³ ì ê é è 0¹æÁÁs1Ö±å¾1ÕÅ¡Æ#yÕÓGÍËIã½¡¿6¡¥³UÆ º ¾ µ ² ²³ Ç ä ¹ Ö µ ±³ ¸ Ç Ì Ê ¹ ½ Á Á ¾ Ö µ ± ¾ ½ ² ² ¸ ââ ââ Ú áââ ß ÚÚ ×Ø¹U¡Õ³UGÇ ÙÚÚ Ì ½ Ö µ ± ¸ Ó ¡ º à Ý Ý Ý ¹ Ö µ ±³ µ Û Æ e¥UXÞ ¹ Ö µ ± Ü · P¥³U® Ý eµ¥¥® ¹ Ö ±³ Ü · eµ¥¥Æ ¹ Ö ±³ µ Û 0ÂVÔUF¸ÓG|¾ Î Ò1D§¥GÐ¡· Î ¾ É ¾ Ç ¾ Ç ¾ ¶ » Ñ ° ¯ Ï ÌÍÊË¹¡É¼È½Ç¡60Â¥½IÁÅ6½§Ä¢¢½¡²!Ã0ÂUIÁG¶6½¡À¿6½¡D§´ ½ ¾ Æ ¾ ¾ ¾ » ¾ ½ ¾ ¶ ¾ ¼ ¶ » ³ ° ¸ ² · ¶ µ ° º8´¥U® ¹ ° ¸ ² · ¶ µ ° ³ ² ± ° ¯ .

¤¨ © 9 9 §§©¨§§§§§§h z 7 ¨ 7 © ¦ ) ¦ h © ! i y @ £ y xC¥¤§ r xG¢¥¤§ r GC@¥¤§ r Gv 2 w v @ £ y x £ y x ! i y @ y pxC v§ r xGv¢¡§ r GC@§ r Gv 2 w @ y x y x u vr sq r ! " §§md§§§§9u§§§9§§§¦§§§G§§§ b a c @ $ ¨ 7 © 7 g 7 T ¦ ¨ c a ¨ ¦ g 7 6 T &|§c b a & b |§a sq r & |` sq r ! b a @ pA§kd¨§§§§i§§§G§§§ $ 7 © 7 9 a ¨ ¦ g 7 6 T ¨ 7 © 7 §§§§9§§§§§§§§§f§§§§§§Ti§9u§§§© 6 T 7 h ¨ ¦ b ¨ ( ¦ © T 7 T 6 ¨ ¨ ) 9 9 !pA§§§CPPy 2 i y y y @ y @ uvr ( 6 ) © §§§ug§§§§§§¨§§§§u§§§§§B 7 T h ¦ c © ¦ ) 8 6 © 7 h ¨ ¦ ¨ u vr y y C§§B w 6 T y uvr ! "Akf& V @ u vr u vr u vr u vr u vr u vr u vr y ww§§¥ 9 y ! ! ` @ $ g 7 T 9 ¦ ¨ c © ¦ 8 "d§§§§§§§§)§6 G§§§ 7 6 T 9 T a §§§©imE§§§f¨7 9 8 6 T z pACHG§H@PCHFGG1§ 3 7§§5h§§§§§§§§§§ 3 ! V @ U @ Q ( @ Q I @ @ D B ¦ 6 & 9 ¦ © 8 ¨ 7 6 5 ¦ 6 T 7 © 8 ¨ 7 6 T §§§§§§§§§§f¨§)§f§§§§¨ tsq ¦ c b r y y y #%1lDB & $ @ @ r ttr r rs r ( ¦ ¨ T b ¨ © z 9 ( ¨ 9 §§§§fD§§)§9 sq r VddUEf& @ @ @ U 7 6 T G§§§ sq r ! Y @ ( W "CH§§1¨§I~xCdU 2 © & i V @ 9 7 6 © 7 9 6 §§§§§§§§hi¨§Ii§7u§§§§h © 7 ¨ c 6 §§§i¨f§ 8 6 T z 7 b z 7 ( §§T ydsq r ! w y x R§Gyd}I & sq r Q( n§Q & ! ¨ 7 Q & 9 R§§)S|Q ¨ 7 ¨ © © D 8 6 T ¦ © h 9 6 T 7 h 6 b ¨ D z §§§§§t§§§§§§f§§§§§)u©9 !x§¨§§§wvp§§7S& 6 6 T Q ¨ ) Q w r r r r r r rs r r r r r r r r r r r r r r ! r tr tr ttsttttrtrttrttrstrttrttrstrttrtsrttrttrtsrttrttrtsr ¨ 6 6 Q ¨ 7 ) u §§§§Tqop§§Q q r sq r sq r sq 9 9 ¨ 6 { h T z §§§§§§7u§§© ksq ¦ ¦ y r 9 ) T ( © §§§u§§§§ 6 © T ( ¨ r sq §§§f§§§i§§§§T 9 ) T ( © 8 6 T ¨ 6 6 §§§i§§§i§§ 9 ) T ( © 8 6 T ¨ 7 ) uvr u vr uvr g 7 T 9 6 ¨ §§§§§§(f§§© 9 9 ! x y & ¨ 6 6 T mlGkj§§§§Q ! x y & ¨ 7 ) RGh§§Q R§§§d& ! y y e ! x y & wG1Y ! y x §§yGw%§§W & ( sq r F T 7 © 9 §§§§§§§§ tsq 6 9 T r 9 ) ¦ ) ( 9 8 6 ) r §§ sq u vr 9 ¨ 7 §§§§¨§§Dt§§5 © © 7 ) D 6 ' @ i h h © pA§§@P§P§H@ §' R§cdR§CRCPCX§§HPC@HGPQ g f @ B e Q @ b a @ b a @ ` @ Y @ ( W @ V U @ 9 1C1§HR§§§§QSR§§HPC@HGEDCA§§§§§§§§§§ 342 @ Q ( @ Q @ ¨ 6 6 T @ ¨ 7 ) Q @ I F @ B @ 9 ¦ © 8 ¨ 7 6 5 ¦ 1§¦§ ' 0 ¨ ) ( & $ #%# ¨ !" ¨§§§§¥ © ¨ ¦ .

©§ ¨ ¤¤ £ ¤ ¤¤ ¤¤ ¢ úûû û ûû ûû ù ¤¤¤ ¤¤ ¤¤ ¢ £ ¡ ¡ ¡ ù ûû ûû ú ûûû ¦ ¡ ¥ ¡ ¤¤ ÿ þý ÿ þ èøý ¡ ÿ øý þ ÿ èþ ý ÿ þ ý ÿ þý ¡ ÿ þý ÿ èþ ý ÿ þøý ÿ èþøý ü £ ¤¤¤ ÿ ¤¤ èþ ý ¢ ÿ þ ý ü ûû ù ú ûûû ûû ÷ øó ò öõ ñ ó ôó ò ñ ëð Ð #ÈÄ µ ª ² Ñ Ð ª µ Æ ¯ ® ¬ pë§§§§§§§« ¿ §Ù§§« × × Õ Õ Ó Ô µ Ï ° ² ¬ ° µ ® ¬ « ² ´ ² Õ î Ó È ¬ ° ® À êE§§§§§§§§§§§§§§§ÕG§§§ Ò Ñ ï Ñ Ð º ¹ "§· §¿ E² Ë §§§§§§§§íÙ§§« ´ µ º « ² ° ® Â Õ Õ À ¿ È ½v}§µ§Ï§°§²§¬°§µ§§¬§«§§²§´§²§Õ§Õ§«²§¬§«§§§§t» Ú » ® ´ á Õ ´ « ì Ñ xÆ â §çÙ§· Â º ¹ Ñ"ë§§§§§§§« ¿ §Æ â À êÔE§§§§f é§§§§§§² â G§§§ Ò Ð ª µ Æ ¯ ® ¬ × × Ó ² Õ µ ¶ ± ² ® ´ « ² ° Ó È ¬ ° ® pA§¯ À º À xÜ RÖ × æè¢À ª§· åÀ ÁxÜ À Ü À §§Û ¾ G§§§§§§² â Ñ Ð ¶ ß Á Ü Ü È ² Õ µ Æ ° À ã ¾ 1A§¯ À º À xÜ RÖ × æÀ §· ¾åÀ xÜ À Ü À §§Û ä¾ G§§§§§§² â ÙÆ â ç Ð ¶ ß Á ª Á Ü Ü ã È ² Õ µ Æ ° Â À ½ » Ï á ® ± « ° « ® ± v|§§§f§§§§²ªi§« ² à ¬ ¬ ª µ ¬ ¬ µ Ø ± ° « §§§§§¼§¬ ß µ §§f§§§§§i§³±§§§§§§§§§t» Ú ª µ ¶ Ï ° µ ¬ Þ ² Ý Ü Ü Û ¶ µ « µ ´ Õ ® ¬ ² ª ª Å º Ñ"ÐH§µ§Æ§§§§§« ª ¯ ® ¬ × × º ¹ Ô º · Å ± µ Ø ² ² Æ Ó È ¬ ° ® ¿ §Ù§· À ÓdE§¹¦§§t§§§G§§§ Ò Ñ "Ðd§§§§§§§« ¿ §×Ð C§¿ Ãª§§§§§§§§§§§§º ª µ Æ ¯ ® ¬ × µ Æ Õ ® ¶ ® ¬ ° ² ® Õ Ö À ¿ È Ô ¯ ¬ « ª Ï Ï Ó È ¬ ° ® À Ód§®§§§§§« ¿ G§§§ Ò Ñ "Ðq§§§§§§§§§§µ À §¯ ËdÀ §¯ À ÉÊE§§µ§§§§§ÄHÃA§· §¿4¾ Ï ´ « Î ² ¬ ° Í ´ ¶ ¶ È Å ± Ç ² ² Æ ° Å Â Á º ¹ Ì À ¿ ½ v»¼§¸§§§³§§§f§§§§§§§³§§f§§§§ª º ¹ · ² ² ¶ ± « µ ´ ´ « ° ² ¬ ¯ ² ± ° « ¯ ® ¬ « .

© ss ss s ss s ss ss ss s ss ss r i p pp uu nuuuuuu ~ w uvUuU uu } yuu u u e p pp t v uuuu p pp pp h p pp pp pp p pp p pp s ss v ss u nu w u u|puy u e uuu ~uu u } u u { z uu u nuuuu ~w uuu Uvu} uu Uuw u u y u u uu u q v ©u ~w uu }bu |u{ zpuxe uuu u u u u u y uuu w u ss r s ss s ss ss ss q e e e e r sss ss ss q e e e h SDDnD©m l d d d l g d d DDd k jih nD©m 6jih 9©D4©fe ©246 lk g d d d pp i ppp pp ppi pp e ppp e h p pp pp l d d d l&D©D& d d d ss r &D©DS© ss l d d d g d ss &D©DS©Vl k fh D©Dd 4fe ©@d 6 s ss l d d d q &D©DS© l k g d d &Ddp9ih 9oDnm x©Ut yw v u r ss ss ss i pp pp q " ## ### # ### # ### ! pp h f e a gFAa `ba ` ` c c ` da ba ` " ## ## ### # ### ## ### # ### ### # ### # ### ### #### ### ### ##" ### # # ### ### ### ### ### ### ### ### ! ! "# ## ### # ### ## ### ! P H ##" QQQ ### QQQ 3 #### QQQQ H #### QQQQ ! QI G "# ### # ### ## ### ! 3 3 3 VWV3 QQ P Q P 3 3 3 QQQ QQQ VWV3 QQQ QQQ 3 3 3 QQQ QQQ VWV3 QQQ QQ VWV3 3 3 3 QQQ I QQQ 3 3 3 QQQ Q P VWV3 QQQ QQ VWV3 3 3 3 QQQ QQQ 3 3 3 QQQ QQQ VWV3 QQQ QQQ 3 3 3 QQQ I VWV3 QQQ QQQQ VWV3 3 3 3 QQQ QQ P 3 3 3 QQQ QQQ VWV3 QQQ QQQ 3 3 3 QQQ QQQ VWV3 QQQ QQ I VWV3 I 3 3 3 3 3 3 YWV3 #" QP QQQ YWV3 ##### 3 3 3 QQQ # 3 3 3 QQQ YWV3 ### QQQ YWV3 ### ! I 3 3 3 3(1F% 4DB3 ) ' E $ 3 ) ' 3 ) ' G (1FE (C H (1@% (E A(1@% (E&R3 H3 ) ' $ 3 3(1F% 4E ) ' 3 G 3 1 )' $ 5 3 ) ' 1 A(@08X6(1F98 (D$ 5 (1@98 (&$ 5 (F0(&$ 3(1FE 4C ) ' 3 3 ) ' 1 3 1 )' % 3 V3 3 3 "# QQ P 3 3 1 ) ' E $ G3 ) ' C H (@% (&7U41@E (S$ #### QQQ V3 QQQ ### H3 ) ' E 3 QQQ 3T6(1F% (S$R3 ### QQQ 3T6(1FE (S$R3 ### G3 ) ' C 3 ! I 3 3 3 VWY3 #" QP QQQ VWY3 ##### 3 3 3 QQQ # 3 3 3 QQQ VWY3 ### QQQ VWY3 ### ! I 3 3 3 V3 3 3 3 #"# QQ P 3 3 1 ) ' E $ 31 ) ' C H (@% (&7GA(@E (&$ #### QQQ V3 QQQ ### 3 1 ) ' E $ 3 QQQ 3 H (F% 4DB3 ### QQQ ### 3 G3 1 ) ' C $ 3 ! I T6(FE 4DB3 H3 1 ) ' E H3 1 'E A(@)% (&$ G 42)E '(D$ 3 1 C P 6(F% (D$B3 QQQ 31 ) ' E $ 3 H (2% (&B3 QQQ 3 QQQ H 3 1 ) ' E G (@0$ 5 4298 '(D$ 3 1 )' 8 3 1 ) 1 QQQ 6(F% (D$B3 QQQ 6(FE (D$B3 5A(@98 (&76420'(&$ 3 1 ) '1 $ 53 1 ) % I G3 1 ) ' C .

© & 9 S .

©Ö Ö Õ Ô SÓ Ì ËÒ Ñ ÑÑ Ñ ÑÑ Ï Ð Ë É Ê ÊÊ Ë Ë È Ê ÊÊ Õ Ô SÓ Ì Ë Ò Ñ ÑÑ Ñ ÑÑ Ï Ð É Ê ÊÊ Ê ÊÊ È Ñ ÑÑ Ñ ÑÑ Ñ ÑÑ Ñ ÑÑ Ñ ÑÑ Ñ ÑÑ ÑÑ ÑÑ Ñ ÑÑ Ñ ÑÑ Ñ ÑÑ Ñ ÑÑ Ñ ÑÑ Ñ ÑÑ ÑÑ Ï Ð Ë Î Í Ì Ë Ì É ÊÊ Ë ÊÊÊ ÊÊ Ë ÊÊÊ Ê ÊÊ ÊÊ ÊÊ Ê ÊÊ Ê ÊÊ Ê ÊÊ Ê ÊÊ Ê ÊÊ Ê ÊÊ ÊÊ È ©¨©© ©© ©© ©©© ©©© ©© ©© ©©© ©©© ©©© ©© ©© ©©© ©©© ©© ©© © ©© § ©© ¦ ¢ ¡ ©©© ¤¥ £ Db Ê ÊÊ ÊÊ ÇÇÇ Æ ÇÇ ÇÇ ÇÇÇ ÇÇÇ ÇÇ ÇÇ ÇÇÇ ÇÇÇ ÇÇ ÇÇÇ ÇÇÇ ÇÇ ÇÇ ÇÇÇ ÇÇÇ ÇÇ ÇÇ ± Ç ÇÇ ± Å 0°¬xªÄ± ²±¯ ® ¼ ¬± ¬ ± ¬ ± ¬ ± ¬ ± ± ± ± ± ±¯ ® ¼ ± ¬ ± ¬ ± ¬ ± ± ± ± ² °¬xªÃ± ¬ ± ± ²± ® « ³m¯¶ ¬µª Â °½ ¬µªf± ±¯ ® ¼ ¬ ¬ ± ¬ ± ± ± ± ¬ ± ± Â± ® ³m¯¶½ ¬¼µª ¿ °®Á ¯Àªf¬± ±¯ ¬ ¬ ± ¬ ± ± ± ± ¬ ± ± ¬ ¬± ¬ ± ²±¯ ® ¬¯ ¬ ³°Á Àªf± ¬ ± ± ± ¬ ± ¬ ¬ ± ± ¬ ¬± ¬ ± ¬± º¹bª ±¯ ®¬¸ ¿¾ ¯ ® ¬ ¼ ¬ |6±m¶½ µªi± ± ± ¬ ¬ ± ± ¬ ¬± ¬ ± ¬± ¿ ¾°®½ ¼b»6¹µª·± ±¯ ¬ ª º ±¯ ®¬¸ ¬ ± ¬ ± ± ± ± ± ± ¬ ± ¬ ± ²±¯ ® ¬« ª 0°x µ´± ± ¬± ± ¬ ± ± ±¯ ¬ « ª ² m¶®x µ´± ¬ ± ¬ ± ± ± ± ²³°®x xª ±¯ ¬« ± ± ± ± ± ± .

ß ß ©Þ Ý ÝÝ ÝÝ ÝÝ Ý ÝÝ Ý ÝÝ Ý ÝÝ Ý ÝÝ Ý ÝÝ Ý ÝÝ ÝÝ ÝÝ Ý ÝÝ Ý ÝÝ Ý ÝÝ ÝÝ Û Ü Ø ÙÙ Ú ÙÙÙ Ù Ù ÙÙ Ù ÙÙ Ù ÙÙ Ù ÙÙ Ù ÙÙ ÙÙ ÙÙ Ù ÙÙ Ù ÙÙ Ù ÙÙ ÙÙ × ÙÙ ÙÙ ÝÝ Ü Ý ÝÝ Ý ÝÝ Ý ÝÝ ÝÝ ÝÝ Ý ÝÝ Ý ÝÝ Ý ÝÝ Ý ÝÝ Ý ÝÝ Ý ÝÝ ÝÝ ÝÝ ÝÝ ÝÝ Û ÙÙ × ÙÙ ÙÙ ÙÙ Ù ÙÙ Ù ÙÙ Ú Ú Ù ÙÙ Ù Ú ÙÙÙ Ù Ù ÙÙ Ù ÙÙ Ù ÙÙ ÙÙ ÙÙØ Ù ÙÙ Ù ÙÙ .

Zerna. and F. D.J. 37(11):1555– 1572. [5] R. Vinolas. McGraw-Hill Inc. [9] A. Classical Mechanics. Englewood Cliffs.E. 2001. Stacke. Cook. Plesha. Holgerson. Distortion of Bearing Rings due to Residual n Stresses. Fritzson. series in dynamics. London. and M. Master’s thesis. Addision-Wesley series in physics. [6] D. Proc. Hindmarsh. 1996. New York. PrenticeHall Inc. Int. J. Cohen and Alan C. Coca˜ o. 101 . Amen House.R. Second edition. CVODE User Guide.pdf. L. Int. Chandrupatla and A. [3] T. 1989. Introduction to Finite Elements in Engineering 2nd ed. Belegundu. Sweden. [2] J.. J.-E. Nordling. Concepts and Applications of Finite Element Analysis 3rd ed. Mase. 1999. Fritzson. BEAST .. Dynamic Simulation of Grinding with Flexibility and Material Removal. 1954. Addision-Wesley Publishing Company.J. Netlib. Carpinteiro and P. John Wiley & Sons. 1994. Time-Domain Dynamic Modelling of the External Plunge Grinding Process. Goldstein.Bibliography [1] K. Nakhimovski. Nieto. 1970. Finite Element Procedures. New Jersey.chalmers. In Proceedings of Multibody Dynamics in Sweeden 2001. Schaum’s outline series. 2001. Continuum mechanics. [12] I. Fung. [11] G.E. Part K. Instn Mech Engrs. Bathe. [7] Y. Malkus.. page http://www.a Rolling Bearing Simulation Tool. 213. Oxford University Press. Theoretical elasticity. Inc.sm. Biera. 1997. D. Prentice-Hall Inc. Green and W.C.. [4] Scott D. and P. 1988.se/MBDSwe Sem01/Pdfs/IakovNakhimovski. Chalmers Univ of Tech. [10] J. Mach. Foundations of solid mechanics. Tools Manufact.S.D.D. 1997. 1965. Prentice Hall. [8] H. and M.

1999. Tonshoff. 20:361–380. 1993.[13] Patrik Nordling. Multibody Systems Dynamics.P. Paul. 2001.P. Engrs. Discretization consideration in moving load ﬁnite element beam models. pages 39–46. [18] R. Inasaki. [15] W.. Finite Elements in Analysis and Design. Sweden. O. Annals of the CIRP. Licentiate thesis. In Proceedings of SIMS 2001. and W. Teukolsky. Lin. Modelling and Simulation of Grinding Process.W. Shabana. Dynamic behaviour of rolling bearings: simulations and experiments. Y. Resonance Conditions and Deformable Body Coordinate Systems. 1997.-H. 192(1):389–398.A. Instn Mech. 21:129–144. 1996. Contributions to a High-level Programming Environment for Scientiﬁc Computing. Shabana. Proc. Rieker. Flexible Multibody Dynamics: Review of Past and Recent Developments. 1992. Annals of the CIRP. [23] L-E.W. 1:189–222. Peters.A. [14] OMG. 2001.Managing Remote Simulations. Linkoping University. Schwertassek.N. 1996. RunBeast. 1976. [21] A. Linkoping University. The Effect of Deformation on the Contact Area in Grinding. S. 215:499–508. Veubeke. [24] S.S Qi. 2001. modeling language speciﬁcation.A. Trethewey. Timoshenko and J. [25] H. The Simulation of Rolling Bearing Dynamics on Parallel Computers. 1987. Licentiate thesis.A. B. Zheng. Journal of Sound and Vibration. Shabana. 1996. Nonlinear Dynamics. 1998. [26] B. Cambridge University Press. Siemers and I.org/uml. [17] W. McGraw Hill Inc. J. Numerical Recipes in C. [22] A. Dynamics of multibody systems 2nd ed. 43(1).B. Cambridge University Press.K. Wallrapp. part J. 2nd edition. Nakhimovski. and H. Vetterling. Press. [16] J. Goodier. 1993. Shabana. and A.F. [19] A. 1994. Sweden.A. [20] A. [27] Lars Viklund.T.omg. Flexible Multibody Simulation and Choice of Shape Functions. and M. The Dynamics of Flexible Bodies. 41(2). 102 . Journal of Engineering Tribology. and T. UML Uniﬁed http://www. Rowe. Morgan. Stacke and D.H. Theory of Elasticity 3rd ed. Flannery.N. 14:895–913. Fritzson.R. M. I. International Journal of Engineering Sciences. H.

Mech. Tools Manufact. Allanson.A.. [31] Xun Chen . McGraw Hill International Edition. Vehicle System Dynamics.. PhD thesis. and Mach. 1998. 30:237–256.C. The Netherlands. Univ. On The Dynamics Of Ball Bearing. Enschede. 1998. of Twente. Int. 22(3):283–304. Mach. Standardization of Flexible Body Modeling in Multibody System Codes.R. D. Struct. 1989. [32] W. B. Roark’s Formulas For Stress & Strain 6th ed. Wallrapp. 103 . Wallrapp. 1994.. W. 1998. [29] O. 38(1-2):41–49. Young. Rowe. Wensing. Part I: Deﬁnition of Standard Input Data.[28] O. [30] J.. Analysis and Simulation of the Grinding Process Part IV: Effects of Wheel Wear. J. Mills.B. Flexible Bodies in Multibody System Codes.

.

general shape functions . The application presents some results from a case where the developed system is used for simulation of flexible ring grinding with material removal. simulation. Nyckelord Keywords flexible. department Datum Date Institutionen för datavetenskap LINKÖPINGS UNIVERSITET Department of Computer and Information Science 2002-12-18 Språk Language Rapporttyp Report category ISBN 91-7373-559-0 Svenska/Swedish X Licentiatavhandling Examensarbete X Engelska/English ISRN LiU-Tek-Lic-2002:62 ISSN C-uppsats D-uppsats Serietitel och serienummer Title of series. numbering 0280-7971 Övrig rapport Linköping Studies in Science and Technology URL för elektronisk version Thesis No. An industrial application example is also described in the thesis. dynamic. mechanics. Additionally. 989 Titel Title Modeling and Simulation of Contacting Flexible Bodies in Multibody Systems Författare Author Iakov Nakhimovski Sammanfattning Abstract This thesis summarizes the equations. multi-body. floating frame of reference. The approach is expected to be computationally less expensive than FEM approaches and easier to use than other reduction techniques. the described technique enables studies of the residual stress release during grinding of flexible bodies. The assumed general shape function approach is also presented. institution Division. elastic.Avdelning. algorithms and design decisions necessary for dynamic simulation of flexible bodies with moving contacts. The overall software system design for a flexible multi-body simulation system BEAST is presented and the specifics of the flexible modeling is specially addressed.

.

Björn Fjellborg: An Approach to Extraction of Pipeline Structures for VLSI High-Level Synthesis. 1992. Patrick Lambrix: Aspects of Version Management of Composite Objects. 1989. Torbjörn Näslund: SLDFA-Resolution . Rolf G Larsson: Aktivitetsbaserad kalkylering i ett nytt ekonomisystem. 1986. Patrick Doherty: A Three-Valued Approach to Non-Monotonic Reasoning. 1989. Yngve Larsson: Dynamic Configuration in a Distributed Environment. Nahid Shahmehri: Affect-Chaining in Program Flow Analysis Applied to Queries of Programs. 1989. Dan Strömberg: Transfer and Distribution of Application Programs. 1987. Chowdhury: Statistical Expert Systems . S-581 11 Linköping. 1993. Tim Hansen: Diagnosing Faults using Knowledge about Malfunctioning Behavior. Torbjörn Näslund: On the Role of Evaluations in Iterative Development of Managerial Support Sytems. Magnus Morin: Predictable Cyclic Computations in Autonomous Systems: A Computational Model and Implementation. Henrik Eriksson: A Study in Domain-Oriented Tool Support for Knowledge Acquisition. 1989. 1992. 1993. Ola Petersson: On Adaptive Sorting in Sequential and Parallel Models. 1993. 1987. Andreas Kågedal: Logic Programming with External Procedures: an Implementation. Magnus Merkel: Temporal Information in Natural Language. 1990. Xinli Gu: Testability Analysis and Improvement in High-Level Synthesis Systems. Johan Ringström: Compiler Generation for Parallel Languages from Denotational Specifications. 1990. 1991. 1989. 1993. 1987. 1988. Staffan Bonnier: Horn Clause Logic with External Procedures: Towards a Theoretical Framework. 1985. 1987. Mats Wirén: On Control Strategies and Incrementality in Unification-Based Chart Parsing. 1984. Ivan Rankin: The Deep Generation of Text in Expert Critiquing Systems. Ulf Cederling: Industrial Software Development . Christer Bäckström: Reasoning about Interdependent Actions. 1989. 1988. 1991. Christos Levcopoulos: New Results about the Approximation Behavior of the Greedy Triangulation. Henrik Nordin: On the Use of Typical Cases for Knowledge-Based Consultation and Teaching.a Special Application Area for Knowledge-Based Computer Methodology. 1987. Michael Jansson: Propagation of Change in an Intelligent Information System. 1989. Olof Johansson: Improving Implementation of Graphical User Interfaces for Object-Oriented KnowledgeBases. Jonas Löwgren: Supporting Design and Management of Expert System User Interfaces. 1993. Simin Nadjm-Tehrani: Contributions to the Declarative Approach to Debugging Prolog Programs. 1990. 1986. 1988. Lars Strömberg: Postmortem Debugging of Distributed Systems. FOA Report B30062E) Arne Jönsson. Kristian Sandahl: Case Studies in Knowledge Acquisition. Lena Srömbäck: Studies in Extended Unification-Based Formalism for Linguistic Description: An Algorithm for Feature Structures with Disjunction and a Proposal for Flexible Systems. Zebo Peng: Steps Towards the Formalization of Designing VLSI Systems. Ulf Nilsson: A Systematic Approach to Abstract Interpretation of Logic Programs. Mikael Pettersson: DML-A Language and System for the Generation of Efficient Compilers from Denotational Specification. Holmes: Using Connectivity Graphs to Support Map-Related Reasoning. Hans Block: SPORT-SORT Sorting Algorithms and Sport Tournaments. Jalal Maleki: ICONStraint. 1993. 1987. Sweden.a Case Study. 1992. Per Österling: Symbolic Modelling of the Dynamic Environments of Autonomous Agents. 1987. Tomas Sokolnicki: Coaching Partial Plans: An Approach to Knowledge-Based Tutoring. 1987.Department of Computer and Information Science Linköpings universitet Linköping Studies in Science and Technology Faculty of Arts and Sciences . 1992. Mikael Patel: An Interactive Flowcharting Technique for Communicating and Realizing Algorithms. Migration and User Acceptance of Expert Systems. Johan Boye: Dependency-based Groudness Analysis of Functional Logic Programs. Christer Hansson: A Prototype System for Logical Reasoning about Time and Action. (Available at: FOA. 1989. Mats Larsson: A Transformational Approach to Formal Digital System Design. 1985. 1989.Licentiate Theses No 17 No 28 No 29 No 48 No 52 No 60 No 71 No 72 No 73 No 74 No 104 No 108 No 111 No 113 No 118 No 126 No 127 No 139 No 140 No 146 No 150 No 165 No 166 No 174 No 177 No 181 No 184 No 187 No 189 No 196 No 197 No 203 No 212 No 230 No 237 No 250 No 253 No 260 No 283 No 298 No 318 No 319 No 326 No 328 No 333 No 335 No 348 No 352 No 371 No 378 No 380 No 381 No 383 No 386 No 398 Vojin Plavsic: Interleaved Processing of Non-Numerical Data Stored on a Cyclic Memory. Ola Strömfors: A Structure Editor for Documents and Programs. 1985. 1984. A Dependency Directed Constraint Maintenance System. Johnny Eckerland: Retargeting of an Incremental Code Generator. 1992. Peter D. 1990. Mehran Noghabai: Evaluation of Strategic Investments in Information Technology. 1989. Box 1165. 1992.Computing Answers for Negative Queries. Rober Bilos: Incremental Scanning and Token-Based Editing. . Johan Hultman: A Software System for Defining and Controlling Actions in a Mechanical System. 1990. 1990. 1988. Peter Åberg: Design of a Multiple View Presentation and Interaction Manager. 1992. Mariam Kamkar. Shamsul I. 1986. Ralph Rönnquist: Network and Lattice Based Approaches to the Representation of Knowledge. 1991. Johan Fagerström: Simulation and Evaluation of Architecture based on Asynchronous Processes. 1992. Jonni Harrius: An Architecture and a Knowledge Representation Model for Expert Critiquing Systems. Tony Larsson: On the Specification and Verification of VLSI Systems. 1993.

Martin Sköld: Active Rules based on Object Relational Queries .utveckling och tillämpning av synsätt och metod. Magnus Werner : Multidatabase Integration using Polymorphic Queries and Views.ett interaktivt perspektiv. Lars Viklund: Contributions to a High-level Programming Environment for a Scientific Computing. 1995. Stefan Svenberg: Structure-Driven Derivation of Inter-Lingual Functor-Argument Trees for Multi-Lingual Generation. 1995. Owen Eriksson: Informationssystem med verksamhetskvalitet . Mikael Lindvall: A Study of Traceability in Object-Oriented Systems Development. 1996. 1994. Ulf Söderman: On Conceptual Modelling of Mode Switching Systems. 1993. 1996. 1994.utvärdering baserat på ett verksamhetsinriktat och samskapande perspektiv. . Göran Forslund: Toward Cooperative Advice-Giving Systems: The Expert Systems Experience. Karin Pettersson: Informationssystemstrukturering. Annika Larsson: Ekonomisk Styrning och Organisatorisk Passion . 1995. Jonas Hallberg: High-Level Synthesis under Local Timing Constraints. Mikael Johansson: Quality Functions for Requirements Engineering Methods. ansvarsfördelning och användarinflytande . Erik Stoy: A Petri Net Based Unified Representation for Hardware/Software Co-Design. Joachim Karlsson: Towards a Strategy for Software Requirements Selection. Kristina Larsen: Förutsättningar och begränsningar för arbete på distans . Mikael Ericsson: Commenting Systems as Design Support—A Wizard-of-Oz Study.No 402 No 406 No 414 No 417 No 436 No 437 No 440 FHS 3/94 FHS 4/94 No 441 No 446 No 450 No 451 No 452 No 455 FHS 5/94 No 462 No 463 No 464 No 469 No 473 No 475 No 476 No 478 FHS 7/95 No 482 No 488 No 489 No 497 No 498 No 503 FHS 8/95 FHS 9/95 No 513 No 517 No 518 No 522 No 538 No 545 No 546 FiF-a 1/96 No 549 No 550 No 557 No 558 No 561 No 563 No 567 No 575 No 576 No 587 No 589 No 591 No 595 No 597 Lars Degerstedt: Tabulated Resolution for Well Founded Semantics. aktiviteter och ansvarsroller. 1995.Efficient Change Monitoring Techniques. Choong-ho Yi: Reasoning about Concurrent Actions in the Trajectory Semantics.A Proposal Based on Active Database Systems. . Anna Moberg: Satellitkontor . 1996. 1995. Lars Poignant: Informationsteknologi och företagsetablering . 1995.En motiv. 1994. 1993. Mikael Lind: Affärsprocessinriktad förändringsanalys .fallstudier av företagsledarutköp ur ett agentteoretiskt perspektiv. 1996. 1995. Henrik Nilsson: A Declarative Approach to Debugging for Lazy Functional Languages. 1996. 1995. Malin Bergvall: Systemförvaltning i praktiken . 1995. Dan Fristedt: Metoder i användning . 1994. 1996. Stefan Cronholm: Varför CASE-verktyg i systemutveckling? . 1995. 1995.En komparativ studie med utgångspunkt i två informationssystemstrategier. 1996. 1997.och konsekvensstudie avseende arbetssätt och arbetsformer. Vadim Engelson: An Approach to Automatic Construction of Graphical User Interfaces for Applications in Scientific Computing. Fredrik Nilsson: Strategi och ekonomisk styrning . 1994.en studie av kommunikationsmönster vid arbete på distans. Camilla Sjöström: Revision och lagreglering .Effekter på produktivitet och region.ett historiskt perspektiv. Peter Jonsson: Complexity of State-Variable Planning under Structural Restrictions. 1994. Peter Carlsson: Separation av företagsledning och finansiering . Patrik Nordling: The Simulation of Rolling Bearing Dynamics on Parallel Computers. 1994. Jörgen Andersson: Bilder av småföretagares ekonomistyrning. Niclas Ohlsson: Software Quality Engineering by Early Identiﬁcation of Fault-Prone Modules.mot förbättring av systemutveckling genom situationell metodkunskap och metodanalys. 1994. 1996. 1994. Pär Carlshamre: A Collaborative Approach to Usability Engineering: Technical Communicators and System Developers in Usability-Oriented Systems Development. 1995. 1996. 1994. 1996. Anders Ekman: Exploration of Polygonal Environments. 1995. 1995. Johan Herber: Environment Support for Building Structured Mathematical Models. 1995. Cecilia Sjöberg: Voices in Design: Argumentation in Participatory Development. 1994. Gustav Fahl: Object Views of Relational Data in Multidatabase Systems. Niclas Andersson: Compilation of Mathematical Models to Parallel Code.Fallstudier i tre byggföretag. Lars Karlsson: Specification and Synthesis of Plans Using the Features and Fluents Framework. 1995. Bo Lagerström: Successiv resultatavräkning av pågående arbeten.en kvalitativ studie avseende centrala begrepp. 1995. Jakob Axelsson: Schedulability-Driven Partitioning of Heterogeneous Real-Time Systems. Eva L Ragnemalm: Towards Student Modelling through Collaborative Dialogue with a Learning Companion.En studie av Sandviks förvärv av Bahco Verktyg. Hans Olsén: Collage Induction: Proving Properties of Logic Programs by Program Synthesis. 1996. Eva Toller: Contributions to Parallel Multiparadigm Languages: Combining Object-Oriented and Rule-Based Programming. Ling Lin: A Value-based Indexing Technique for Time Sequences. Peter Loborg: Error Recovery Support in Manufacturing Control Systems. Niclas Wahllöf: A Default Extension to Description Logics and its Applications. Anders Avdic: Arbetsintegrerad systemutveckling med kalkylkprogram. Jonas Lind: Creditor . 1994. 1996. 1996. 1994. 1994. 1996. 1997. 1995.erfarenheter från fyra svenska företag. 1994. 1996. 1994. Hee-Cheol Kim: Prediction and Postdiction under Uncertainty. 1996. Esa Falkenroth: Data Management in Control Applications . 1994. Jörgen Lindström: Chefers användning av kommunikationsteknik. Johan Jenvald: Simulation and Data Collection in Battle Training. Staffan Flodin: Efficient Management of Object-Oriented Queries with Late Binding.Firm Relations: an Interdisciplinary Analysis.

Ett stöd för multiflex? 1997. 1998. Carl-Johan Westin: Informationsförsörjning: en fråga om ansvar .egenskaper.Kreditriskhantering ur ett agentteoretiskt perspektiv. 2000. Anders Ferntoft: Elektronisk affärskommunikation . 1998. Ulf Melin: Informationssystem vid ökad affärs. Ulf Seigerroth: Integration av förändringsmetoder . Charlotte Björkegren: Learning for the next project . Tommy Wedlund: Att skapa en företagsanpassad systemutvecklingsmodell . Patrik Hägglund: Programming Languages for Computer Algebra. 1998. 1997. 1997. Carita Åbom: Videomötesteknik i olika affärssituationer . 1998. Henrik André-Jönsson: Indexing time-series data using text indexing methods. Howard: Designing dynamic visualizations of temporal data. 1998.möjligheter och hinder. Peter Ingels: A Robust Text Processing Technique Applied to Lexical Error Recovery. Jonas Mellin: Predictable Event Monitoring. 1998. Analys och värdering av systemutvecklingsmodeller och dess användning. Erik Larsson: High-Level Testability Analysis and Enhancement Techniques.1999. 1998. Anders Bäckström: Värdeskapande kreditgivning . . Martin V. Mats Gustafsson: Bringing Role-Based Access Control to Distributed Systems. 1998.No 598 No 599 No 607 No 609 FiF-a 4 FiF-a 6 No 615 No 623 No 626 No 627 No 629 No 631 No 639 No 640 No 643 No 653 FiF-a 13 No 674 No 676 No 668 No 675 FiF-a 14 No 695 No 700 FiF-a 16 No 712 No 719 No 723 No 725 No 730 No 731 No 733 No 734 FiF-a 21 FiF-a 22 No 737 No 738 FiF-a 25 No 742 No 748 No 751 No 752 No 753 No 754 No 766 No 769 No 775 FiF-a 30 No 787 No 788 No 790 No 791 No 800 No 807 Rego Granlund: C3Fire .En studie av fyra revisionsbyråer. Per-Ove Zetterlund: Normering av svensk redovisning . 1997.en del i företags styrning. 1998. Jens-Olof Lindh: Analysing Traffic Safety from a Case-Based Reasoning Perspective. Smalltalk . Fredrik Öberg: Object-Oriented Frameworks . 1999. Tim Heyer: COMPASS: Introduction of Formal Methods in Code Development and Inspection.aktiviteter och uppdrag i fem stora svenska organisationers operativa informationsförsörjning. 1999. 1998.a suitable Real-Time Language.A Microworld Supporting Emergency Management Training. 1997. 1999. Man Lin: Formal Analysis of Reactive Rule-based Programs. Jo Skåmedal: Arbete på distans och arbetsformens påverkan på resor och resmönster. 1998. 1999. 1999.Bearers and barriers in knowledge transfer within an organisation. Juha Takkinen: CAFE: Towards a Conceptual Model for Information Management in Electronic Mail. 1999. Åse Jansson: Miljöhänsyn . värdering och vidareutveckling i T50-bolag inom ABB. Pawel Pietrzak: Static Incorrectness Diagnosis of CLP (FD). 1998.genom rekonstruktion. Boris Karlsson: Metodanalys för förståelse och utveckling av systemutvecklingsverksamhet. 1998. David Byers: Towards Estimating Software Testability Using Static Analysis. 1999. 1997. 2000. 1997. Jonas S Karlsson: A Scalable Data Structure for a Parallel Data Server. Marie-Therese Christiansson: Inter-organistorisk verksamhetsutveckling . Klas Gäre: Verksamhetsförändringar i samband med IS-införande. 1997.Regression and Tractability.En teori om aktörsarenor i samverkan om utbyte av information. 1999. 1999. Christina Wennestam: Information om immateriella resurser.en modell för välgrundad metodintegration. 1997. 1999. 1999. 1998. 1997.och processorientering . Joakim Gustafsson: Extending Temporal Action Logic for Ramification and Concurrency. Gunilla Ivefors: Krigsspel coh Informationsteknik inför en oförutsägbar framtid. 1998. 1997. Ågerfalk: Pragmatization of Information Systems . 1997 Jukka Mäki-Turja:. 1998.A New Strategy for Case Tool Development. Anders Subotic: Software Quality Inspection. 1997. 1998. 1999. 1997. 1997. Thomas Padron-McCarthy: Performance-Polymorphic Declarative Queries.en studie av projektledning i tre installations. strategier och utveckling.A Theoretical and Methodological Outline. Svein Bergum: Managerial communication in telework. Silvia Coradeschi: A Decision-Mechanism for Reactive and Coordinated Agents. Per-Arne Persson: Toward a Grounded Theory for Support of Command and Control in Military Coalitions. Magnus Lindahl: Bankens villkor i låneavtal vid kreditgivning till högt belånade företagsförvärv: En studie ur ett agentteoretiskt perspektiv. Jan Håkegård: Hiera rchical Test Architecture and Board-Level Test Controller Synthesis.En studie av tillkomsten av Redovisningsrådets rekommendation om koncernredovisning (RR01:91). Joakim Eriksson: Specifying and Managing Rules in an Active Real-Time Database System. Jimmy Tjäder: Projektledaren & planen . 1999. 1998. 2000. Jan Ollinen: Det flexibla kontorets utveckling på Digital . 1999. En studie av berättelser om möten. Marcus Bjäreland: Two Aspects of Automating Logics of Action and Change . Johan Alvehus: Mötets metaforer. Fredrik Eklund: Declarative Error Diagnosis of GAPLog Programs.kontaktkostnader och kontaktprocesser mellan kunder och leverantörer på producentmarknader. Jesper Andersson: Towards Reactive Software Architectures. Investeringar i forskning och utveckling samt i personal inom skogsindustrin. 1997.metoder som stöd vid utveckling av partnerskap och informationssystem. Håkan Nilsson: Informationsteknik som drivkraft i granskningsprocessen . Bengt E W Andersson: Samverkande informationssystem mellan aktörer i offentliga åtaganden . Pär J. 1998. Anders Henriksson: Unique kernel diagnosis.och systemutvecklingsprojekt. 1997. Tobias Ritzau: Real-Time Reference Counting in RT-Java. Erik Berglund: Use-Oriented Documentation in Software Development.

Ett koncernperspektiv. 2001. Benneth Christiansson: Att komponentbasera informationssystem . Niklas Sandell: Redovisning i skuggan av en bankkris . 2002. 2002. 2002. Gert Jervan: High-Level Test Generation and Built-In Self-Test Techniques for Digital Systems. 2001. 854 No 863 No 881 No 882 No 890 Fif-a 47 No 894 No 906 No 917 No 916 Fif-a-49 Fif-a-51 No 919 No 915 No 931 No 933 No 938 No 942 No 956 FiF-a 58 No 964 No 973 No 958 Fif-a 61 No 985 No 982 No 989 Flavius Gruian: Energy-Aware Design of Digital Systems. Henrik Lindberg: Webbaserade affärsprocesser . Sorin Manolache: Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times. Magnus Kald: The role of management control systems in strategic business units. Ola Pettersson: Deliberation in a Mobile Robot. 2002. Lars Hult: Publika Gränsytor . mentala modeller. . Emma Hansson: Optionsprogram för anställda . 2000. Stefan Svarén: Styrning av investeringar i divisionaliserade företag . 2000. Dan Lawesson: Towards Behavioral Model Fault Isolation for Object Oriented Control Systems.en studie av svenska börsföretag. 2000. 2000. Johan Moe: Execution Tracing of Large Distributed Systems. 2000.Erfarenheter från ett FOU-samarbete. Patrik Haslum: Prediction as a Knowledge Representation Problem: A Case Study in Model Design.a Case Study of Pharmaceutical Drug Development. Göran Hultgren: Nätverksinriktad Förändringsanalys . 2001. Robert Sevenius: On the instruments of governance . 2001. Fredrik Karlsson: Meta-Method for Method Configuration .En studie om teknikens aktiva och passiva roll i en organisation.A Rational Unified Process Case. 2002. 2001. en studie av datorimplementerade modeller av verksamhet som stöd för beslut om anskaffning av JAS 1982.A law & economics study of capital instruments in limited liability companies. 2002.Värdering av fastigheter.Vad säger teori och praktik?. Iakov Nakhimovski: Modeling and Simulation of Contacting Flexible Bodies in Multibody Systems.Möjligheter och begränsningar. 2000. 2002. 2002. 2002. Stefan Holgersson: IT-system och filtrering av verksamhetskunskap . Ewa Braf: Organisationers kunskapsverksamheter . 2000. Luis Alejandro Cortes: A Petri Net Based Modeling and Verification Technique for Real-Time Embedded Systems. Peter Bunus: Debugging and Structural Analysis of Declarative Equation-Based Languages. Per Oscarsson:Informationssäkerhet i verksamheter . 2002.perspektiv och metoder som stöd för förståelse och utveckling av affärsrelationer och informationssystem.No 809 FiF-a 32 No 808 No 820 No 823 No 832 FiF-a 34 No 842 No 844 FiF-a 37 FiF-a 40 FiF-a 41 No. Peter Aronsson: Automatic Parallelization of Simulation Code from Equation Based Simulation Languages. 2000. Johan Petersson: Lokala elektroniska marknadsplatser .en kritisk studie av ”knowledge management”. Karin Hedström: Kunskapsanvändning och kunskapsutveckling hos verksamhetskonsulter . 2000. 2000. 2000. Paul Pop: Scheduling and Communication Synthesis for Distributed Real-Time Systems.kvalitetsproblem vid analyser och beslutsfattande som bygger på uppgifter hämtade från polisens IT-system. 2002. 2001. 2001. Linda Askenäs: Affärssystemet .informationssystem för platsbundna affärer. Fredrik Elg: Ett dynamiskt perspektiv på individuella skillnader av heuristisk kompetens. Diana Szentiványi: Performance and Availability Trade-offs in Fault-Tolerant Middleware.ett designexempel. Susanne Odar: IT som stöd för strategiska beslut. 2002. Bourhane Kadmiry: Fuzzy Control of Unmanned Helicopter. 2000. 2002. 2002. 2000. 2001. 2001. mål och konfidens i kontroll av mikrovärlden Moro.begrepp och modeller som stöd för förståelse av informationssäkerhet och dess hantering. Yuxiao Zhao: XML-based Frameworks for Internet Commerce and an Implementation of B2B e-procurement. Mikael Cäker: Vad kostar kunden? Modeller för intern redovisning. Per-Arne Segerkvist: Webbaserade imaginära organisationers samverkansformer. Annika Flycht-Eriksson: Domain Knowledge Management inInformation-providing Dialogue systems. 2001. intelligens. 2000. Fredrika Berglund: Management Control and Strategy . Lin Han: Secure and Scalable E-Service Software Delivery. 2001. Jean Paul Meynard: Control of industrial robots through high-level task programming.

Sign up to vote on this title

UsefulNot useful- A Motion Cueing Model for Mining and Forestry Simulator Platforms Based on Model Predictive Control
- 00093068
- ansys
- macqueen_thesis.pdf
- spe_27051_pa.pdf
- Macqueen Thesis
- Implementation of Accelerated Models EMTP
- Elmer Lut Dec 2011 Ra Back
- 2014 Mbrc Simulation RT Control
- A New Grouping Measure for Evaluation of Machine-Component Matric.pdf
- CASTEM FINITE ELEMENT MODELLING
- Modular Dynamical Simulation of Mechatronic and Coupled Systems
- Paper_627
- Briefing Slide Dp1
- Simulacao Dinamica e o Engenheiro de Processos
- Vittal Rao Lecture 1
- G. K. Costa Et Al
- 7CEM2425 Dynamic Systems & Simulation - Lab 2 Report (Michael M. Wijetunge de Silva)
- Simulation
- Intro_PS_nu
- Fluent and CFX Tascflow
- OJS_file
- circular-accelerators-Amundson.pdf
- APLUS 121 Getting Started EO Modeling
- Medici
- Expertise
- mat341
- Random Walk
- Implicit and Explicit FE Simulation of Soft-pad Griding of Silicon Wafer
- EXTENDED FINITE STRIP METHOD FOR PRISMATIC PLATE AND SHELL STRUCTURES.pdf
- Full Text 01

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue reading from where you left off, or restart the preview.

scribd